thomas olsson creating, managing and ...olsson/diplomityo_thomas_olsson_sharme.pdfthomas olsson...

THOMAS OLSSON CREATING, MANAGING AND SHARING MEMORIES WITH MOBILE PHONES: A USER-CENTERED DESIGN APPROACH Master of Science Thesis

Examiner: Professor Kaisa Väänänen-Vainio-Mattila Examiner and topic approved in the Information Technology Department Council meeting on 16 August 2006

I

ABSTRACT TAMPERE UNIVERSITY OF TECHNOLOGY Master’s Degree Programme in Information Technology OLSSON, THOMAS: Creating, Managing and Sharing Memories with Mobile Phones: A User-Centered Design Approach Master of Science Thesis, 111 pages, 7 Appendix pages May 2007 Major: Usability engineering Examiner: Professor Kaisa Väänänen-Vainio-Mattila Keywords: usability, user-centered design, personal content management, mobile application, context awareness With the recent rapid development of mobile phones and various capturing devices, capturing and sharing photos, videos, messages and other memories have become a remarkable trend. To design successful systems for personal memory management, the end-users have to be taken into account. This thesis work describes the user-centered design (UCD) process of a prototype for personal memory management with mobile phones. The research and prototype design was steered by a futuristic concept called SharMe – developed by Nokia Research Center. The purpose of the study was to introduce the UCD approach in designing a new mobile application. After a benchmark of existing related applications, contextual inquiries as part of Contextual Design were conducted. Camera phone and Flickr users were inquired to reveal the current activity with their digital memories. Affinity diagram and sequence and interaction models were composed to consolidate the user data. In addition, design drivers for new content management systems were formulated. Based on the user study results, the design phase of a new mobile prototype for memory management followed. The prototype design process was iterative with high user involvement. Paper prototype tests were conducted with both a low detail paper prototype in the early design phase, and later on with more detailed mock-up screens and menus. After intensive paper prototyping and redesigning the prototype was ready to be evaluated with end-users. The resulted prototype is an application for mobile phones for capturing, managing, browsing and sharing users’ digital memories. It involves various novel ideas, such as event-based managing and sharing of memories, buffer capturing and context awareness. As the focus was to design the prototype for the existing needs of the users, the prototype is slightly futuristic by supposing certain technical improvements in modern mobile phones. Acceptability evaluation of the prototype was conducted continuously as a part of the paper prototype tests, but also as a separate test round. According to the users, event-based approach in memory management was regarded very useful and intuitive. However, the buffer capturing feature polarized users as its usefulness and effectiveness was questioned. Continuous metadata capturing received a warm welcome. Nevertheless, to be able to evaluate the acceptability of such prototype, implementation is and longitudinal studies of the usage are required.

II

TIIVISTELMÄ TAMPEREEN TEKNILLINEN YLIOPISTO Tietotekniikan koulutusohjelma OLSSON, THOMAS: Käyttäjän Muistojen Tallentamiseen, Hallintaan ja Jakamiseen Tarkoitetun Mobiilisovelluksen Käyttäjäkeskeinen Suunnittelu Diplomityö, 111 sivua, 7 liitesivua Toukokuu 2007 Pääaine: Käytettävyys Tarkastaja: Professori Kaisa Väänänen-Vainio-Mattila Avainsanat: käytettävyys, käyttäjäkeskeinen suunnittelu, henkilökohtaisten muistojen hallinta, mobiilisovelluksen suunnittelu, kontekstitietoisuus Mobiilipuhelinten ja tallennusteknologian kehityksen myötä myös kuvien, videoiden, viestien ja muiden muistojen tallentaminen ja jakaminen on noussut käyttäjien suosioon. Onnistuneiden järjestelmien suunnittelussa on erittäin tärkeää ottaa myös käyttäjät huomioon. Tämä diplomityö kuvaa ihmisten digitaalisten muistojen käyttöön tarkoitetun mobiilijärjestelmän prototyypin käyttäjäkeskeisen suunnitteluprosessin. Prototyypin suunnittelu ja siihen liittyvä tutkimus sai aineksia tulevaisuuteen suuntautuneesta SharMe – konseptista, jonka on luonut Nokian tutkimuskeskus (NRC). Työn tarkoituksena on tuoda käyttäjäkeskeisen suunnittelun lähestymistapa osaksi uuden matkapuhelinsovelluksen suunnittelua. Olemassa olevien samantyyppisten järjestelmien arvioinnin jälkeen suoritettiin contextual inquiryjä osana Contextual Design suunnittelumallia. Kamerapuhelimen sekä Flickr:n käyttäjiä tarkkailtiin ja haastateltiin nähdäksemme käyttäjien nykyistä toimintaa muistojen kanssa. Samankaltaisuusseinä sekä sekvenssi- ja vuorovaikutusmalleja laadittiin käyttäjädatan yhdistämiseksi. Näiden pohjalta kirjoitettiin ohjenuoria suunnittelulle. Käyttäjätutkimusaineiston keruuseen ja analysointiin pohjautuen seurasi uuden mobiiliin muistojen hallintaan tarkoitetun prototyypin suunnittelu.. Suunnitteluprosessi oli erittäin iteratiivinen ja käyttäjät osallistuivat runsaasti prototyypin suunnittelun evaluointiin. Paperiprototyyppitestejä suoritettiin sekä alkuvaiheessa hyvin matalan tarkkuustason prototyypillä että suunnittelun loppuvaiheessa tarkemmalla ja graafisemmalla prototyypillä. Intensiivisen paperiprototypoinnin jälkeen prototyyppi oli valmis hyväksyttävyystestausta varten. Lopullinen prototyyppi on matkapuhelinsovellus digitaalisten muistojen tallennukseen, käsittelyyn, selailuun ja jakamiseen. Prototyyppi sisältää useita uutuustekijöitä, kuten tapahtumapohjaisen toimintamallin, puskuritallennustoiminnon ja kontekstitietoisuutta. Sovellus on lievästi futuristinen olettaen matkapuhelimilta tiettyjä teknologisia parannuksia nykyisiin verrattuna. Tarkoituksena kun oli suunnitella prototyyppi nimenomaan käyttäjien nykyisiin tarpeisiin keskittyen. Prototyypin hyväksyttävyystestausta suoritettiin sekä jatkuvasti paperiprototyyppitestien aikana että erillisenä testikierroksenaan. Tapahtumapohjainen toimintamalli nähtiin erittäin hyödyllisenä, kun taas puskuritallennus jakoi mielipiteitä. Sen tehokkuutta toimintatarkoituksessaan hieman epäiltiin. Pitää huomata että, kunnollinen evaluointi vaatisi toteutetun järjestelmän käyttöä, ja mielellään pitkällä aikajänteellä.

III

PREFACE This thesis was written during spring 2007 in cooperation between Nokia Research Group and Institute of Human-Centered Technology in Tampere University of Technology. In TUT/IHTE, the thesis project was conducted with a five-researcher group. My contribution to this work – together with colleagues – was the user-centered design process and the designed prototype for life memory management. I could not have imagined a better topic for my thesis. After studying such a wide field there are plenty of options to focus on in the future. Conducting an extensive user-centered design process and having the opportunity to design a new system was both highly intriguing and a chastening experience. I am grateful for the freedom in aiming the study and being trusted in the choices I made. It would not be totally incorrect to say that I have learned more about usability and user-centered design during the project than by the time I started it. While the empiric part of the study was the most pleasant, the writing process was the most challenging and exhausting part of the thesis. The greatest challenges were related to the vast amount of user data and the challenges in interpretation when having to make compromises. The study was a multidimensional research process of which, because of the wide perspective, one would not have been able to manage by oneself. Hereby, several people deserve acknowledgements. First and foremost, I wish to thank my associate Minna Wäljas without whom the work would be only half of this. I wish to thank our project manager Marika Lehtonen and colleague Hannu Soronen for joyous teamwork and the captivating perspective during the project. Colleagues at Nokia Research Center, Dana Pavel, Pertti Huuskonen, Salla Myllylä and Markku Laitkorpi deserve many thanks for the interesting research topic, and fluent and inspirational collaboration. Professor Kaisa Väänänen-Vainio-Mattila’s input to the work has been significant and I want to thank her for being a tireless encourager and teaching uncountable matters. Last but not least, I wish to express my gratitude to my belowed Jaana for putting up with me during the laborious writing process, and being there when needed. Thank you cats, Väinö, Ronja, Marta and Ninni, for bringing joy in my life. Thank you Mom, Dad, Nicklas and Linda. You are the best. Tampere, May 18th, 2007 Thomas Olsson

IV

CONTENTS Abbreviations ................................................................................................................ VII 1. Introduction...............................................................................................................1

1.1. Background .......................................................................................................1 1.2. Objectives and Methodology ............................................................................3 1.3. Structure of the Thesis ......................................................................................4

2. SharMe Concept........................................................................................................6 3. Personal Content Management (PCM) .....................................................................9

3.1. Defining Personal Content Management ..........................................................9 3.2. The Nature of Memory....................................................................................11

3.2.1. Operational Principles of the Human Memory .......................................11 3.2.2. The Concept of Memory .........................................................................11 3.2.3. Recalling Memories ................................................................................12

3.3. User Activity around Personal Content ..........................................................13 3.3.1. Capturing Content ...................................................................................13 3.3.2. The Usage Purposes of Content ..............................................................14 3.3.3. Sharing Content.......................................................................................16 3.3.4. Organizing and Processing Content ........................................................18 3.3.5. Retrieving, Browsing and Searching Content.........................................19

3.4. Categorizing the Existing PCM Applications.................................................20 3.5. Recapitulating Personal Content Management ...............................................21

4. User-Centered Design .............................................................................................22 4.1. Background .....................................................................................................22 4.2. Defining User-Centered Design......................................................................23 4.3. User-Centered Research Methods...................................................................25

4.3.1. User Data Gathering Methods.................................................................26 4.3.2. User Data Analysis..................................................................................28 4.3.3. Designing the New Product ....................................................................29 4.3.4. Evaluation Methods ................................................................................29 4.3.5. Design Processes Summarized................................................................30

4.4. Contextual Design...........................................................................................31 4.4.1. Contextual Inquiries ................................................................................32 4.4.2. Work Modeling .......................................................................................33 4.4.3. User Data Consolidation .........................................................................33 4.4.4. Work Redesign........................................................................................34 4.4.5. User Environment Design .......................................................................34 4.4.6. Mockup and Testing................................................................................34 4.4.7. Putting into Practice ................................................................................34 4.4.8. Contextual Design Revisited...................................................................35

4.5. UCD Summarized ...........................................................................................35 5. Context Awareness in PCM Systems......................................................................36

V

5.1. Defining Context Awareness ..........................................................................36 5.1.1. Definition of Context ..............................................................................36 5.1.2. Definition of Context Awareness............................................................39 5.1.3. Mobile Context .......................................................................................40

5.2. Utilizing Context Awareness ..........................................................................41 5.2.1. Utilizing Location Awareness.................................................................42 5.2.2. Entering a Context ..................................................................................42 5.2.3. Using Context Data as Metadata.............................................................43 5.2.4. The Challenges of Context Awareness ...................................................46

6. User-Centered Design in SharMe Project ...............................................................49 6.1. The User-Centered Design Process in SharMe...............................................49

6.1.1. The Goals of the Process.........................................................................49 6.1.2. UCD Methodology..................................................................................50 6.1.3. Focusing the Prototype............................................................................52

6.2. Benchmarking Personal Content Management Applications .........................53 6.3. User Needs Research.......................................................................................54 6.4. Applied Contextual Design.............................................................................55

6.4.1. Contextual Inquiries ................................................................................56 6.4.2. User Data Consolidation .........................................................................59 6.4.3. Work Redesign........................................................................................63 6.4.4. Paper Prototyping and Evaluation...........................................................64

6.5. Revealing Further User Needs with the Help of the New Prototype ..............67 6.6. Summary of the Conducted UCD Process ......................................................68

7. Results of the User-Centered Design ......................................................................69 7.1. Related PCM Applications Benchmark ..........................................................69

7.1.1. Flickr .......................................................................................................69 7.1.2. Nokia Lifeblog ........................................................................................71 7.1.3. Context Watcher......................................................................................71 7.1.4. Meaning – Merkitys ................................................................................72 7.1.5. Multiply...................................................................................................73 7.1.6. Conclusions of the Benchmark ...............................................................73

7.2. User Needs ......................................................................................................74 7.3. User Behavior and Habits ...............................................................................75

7.3.1. Events for Close Group of People...........................................................75 7.3.2. The Nature of Viewing............................................................................76 7.3.3. The Differences between Camera Phone Users and Flickr Users...........76 7.3.4. Summary of the Consolidated Models....................................................77

7.4. Design Drivers ................................................................................................78 7.5. Resulted Prototype ..........................................................................................81

7.5.1. Event-Based Approach ...........................................................................83 7.5.2. Browsing the Memories ..........................................................................85 7.5.3. Annotating, Tagging and Rating .............................................................87

VI

7.5.4. Continuous Capturing .............................................................................88 7.6. Acceptability of the Prototype ........................................................................90 7.7. Design Ideas for the Future .............................................................................92 7.8. Summary of the Results ..................................................................................94

8. Conclusions.............................................................................................................95 8.1. Summary of the Thesis....................................................................................95 8.2. Discussion .......................................................................................................96 8.3. Future Work ..................................................................................................100

References .....................................................................................................................102 Web References ............................................................................................................109 Internal References........................................................................................................111 Appendix A. Summarized Question Outline from Flickr Contextual Inquiries Appendix B. Consolidated Models from Contextual Inquiries

VII

ABBREVIATIONS 3G The third generation mobile phones with high connectivity API Application Programming Interface Bluetooth A data transfer protocol widely used in mobile devices Blog Web log. A user-generated website, usually with journalistic

approach CD Contextual Design CI Contextual Inquiry Cell ID Information on which base transceiver stations’ range the mobile

phone is in. EXIF Exchangeable Image File Format GPS Global Positioning Service HCI Human-Computer Interaction ISO International Standards Organization IT Information Technology MMS Multimedia Messaging Service PC Personal Computer PCM Personal Content Management PDA Personal Digital Assistant POI Point of Interest S60 Nokia’s current standard software platform for mobile phones. SharMe A futuristic concept made by Nokia Research Center TEKES Finnish Funding Agency for Technology and Innovation

(Teknologian ja innovaatioiden kehittämiskeskus) UCD User-Centered Design UED User Environment Design UI User Interface UML Unified Modeling Language Web2.0 The second generation of web-based services, such as social

networking and communication tools WLAN Wireless Local Area Network WWW World Wide Web

1

1. INTRODUCTION

This thesis is a part of the SharMe project conducted in cooperation between Nokia Research Center and Institute of Human-Centered Technology in Tampere University of Technology in the beginning of 2006. SharMe is a research project about creating, managing and sharing user memories with mobile phones. SharMe includes a concept for capturing, managing and sharing memories which in this study was approached with a user-centered manner. The initial SharMe concept is introduced in Chapter 2.

1.1. Background

Through ages people have had an inherent need to share their memories with other people. The memories have been verbal stories and traditions that have lived long in various communities. Technology has enabled capturing the memories almost as vivid as they were when they occurred. For the past century, capturing memories has been mostly photos organized in albums, perhaps with some textual information attached. The development of recording and data transfer technology, and the decreased cost of storing and sharing media files has made it possible for people to record most of the moments they consider necessary. This has led to vast digitalization where information about people is stored in various ways in, for example, one’s own photos, public writings, consumer and authority registers, and other people’s digital memories (Lehikoinen et al 2007). To control this rapidly grown amount of user-related data, efficient and usable systems for personal content management are required. Memories are frequently exchanged with other people in small communities, such as groups of friends, family or some other group with common interest (Sarvas et al. 2005). Often, sharing is the very reason for capturing in the first place. Hence, the aspect of sharing memories is of utmost importance in content management. Traditionally sharing has been mostly viewing the photos or other content together and spicing them with stories and anecdotes. With the recent mushrooming of www-based photo and audio sharing and blogging sites, dealing with life memories has become one of the most predominant pastime for users in the Internet. The applications have enabled totally new forms of user-generated content management, such as sharing photos publicly, viewing the files from various device user interfaces and annotating other people’s photos and videos. According to a study by Sarvas et al. (2005), 89 percent of digital photos are shared at least once and most of the sharing is done within three days of being taken. Publishing can now happen anywhere, anytime.

1. Introduction 2

Being so subjective, the imperfect sensory organs of humans distort the truth and focus only on narrow parts of the environment. Moreover, the memories formed in the human memory are biased with past experiences (Sinkkonen et al. 2006). To enrich the memories and make them more realistic, various elements from the context may be captured (Dey 2001). Instead of partial information the users would have full access to all the information the device’s sensors are able to capture. The information about context can be illustrated as distinct elements in the memory (e.g. charts or maps) or be stored as textual tags, for example to serve as keywords in searching. Capturing context information ensures the users’ to be able to associate their memories with various aspects in the experienced event and thus remembering it more easily and clearly. Because of non-immediate availability of capturing devices some of the important moments are missed (Gemmel et al. 2002). New application concepts with pervasive and continuous capturing could be able to make it easier for users to capture all the memories regarded interesting. Consequently, constant availability of capturing devices and utilizing their potential are ideas whose detailed implementations have to be studied with users. In order to understand what and when it is appropriate to record, and what is desired to be kept and shared, extensive user studies about their needs, requirements, fears and ethics have to be performed. Moreover, some of the privacy and ethical issues are regulated with laws. Recent rapid development in the mobile phone branch has produced devices capable of capturing high quality visual and audio data. Enhanced processing power has enabled consumption of more demanding media on mobile phones, such as photos, music and video. Recently, the mobile phone screens have grown slowly and the resolution and color quality has got better and better. The 3G generation data transfer, grown storage capacity and enhanced battery technology have made it possible to capture and share recordings anywhere and anytime, and store them either locally in the device or remotely in a network. Sharing can be done as enclosed with certain users or publicly in, for example, www-galleries. Having so many possibilities integrated in one device makes the mobile phone most dexterous platform for capturing, viewing and sharing memories. (Sarvas 2006) Mobile phone cameras are currently the most sold cameras in the world (Sarvas 2006). Event though integrated in mobile phones they are normally thought as cameras, which gives the mobile phone’s camera a slightly negative image, because the technology is currently not at the same level than in digital cameras (Aaltonen 2007). Designing more diverse functionalities, such as context awareness and easier sharing paradigms, in the mobile phones could allow them more foothold in the field of content management. Being practically always carried along, camera phones can be also called as ubiquitous devices (Sarvas 2006). This can be both an advantage and a disadvantage. “Although

1. Introduction 3

they are the most practical devices to capture and share memories they are also the most vulnerable to losing, stealing, or just wear-and-tear” (Pavel 2006). The mushrooming of mobile cameras has changed the general attitudes towards being photographed. With the pervasive capturing devices the people have got used to the constant loss of privacy. Even though the rapid growth of mobile capturing, the mobile activities around users’ self-created content are still shaping. Furthermore, introducing technology just for the sake of technology is risky considering the success of the products. User-centered design is an essential approach when dealing with such personal and private matters as memories are. Users must have control over how the memories are captured, used and shared. Taking ethics into account, also the privacy of the near-by people must be considered when recording.

1.2. Objectives and Methodology

This study was motivated by partly the interest in users’ needs related to future personal memory management systems, and partly by developing new concepts and innovations for personal memory management with mobile phones. On one hand, this was design research as a constructive study, and on the other hand, a study trying to reveal and explain certain phenomena in the usage of users’ memories. The role of the author in the research process was significant. The benchmarking (see Section 6.2) was solely performed by the author. In addition, the author contributed also to the used needs study (see Section 6.3). Although conducted as a working pair, the UCD process and prototype design and evaluation (see Section 6.4) was for the most part performed by the author. The main objective of the study was to bring the end-users’ perspective to the development of the SharMe concept. This was attained by extensive user needs research and research on users’ current activity with personal digital memories. The focus of the research was in both users’ needs, requirements, fears and suspicions, and revealing the current interaction, sequences and patterns with users’ memories. However, the ultimate goal was to design a prototype taking these matters into account and evaluating it with target group users. The existing SharMe concept was to be concretized with user studies and reconciling users’ needs in the concept. As the SharMe concept is aimed at the future, in the research process and in analyzing the results had to be paid attention to the future aspect in them, too. The goals of UCD are elaborated in Section 6.1.1. The contribution of this thesis is a new prototype innovation and knowledge of the design and evaluation phases of the prototype. There again, the prototype can also be seen as a remarkable enhancement of an existing system. In the classification of Järvinen & Järvinen (2004), the new prototype can be categorized as automating,

1. Introduction 4

humans’ abilities broadening, entertaining and creating art experiences. The thesis only addresses to the design of the prototype – not the implementation. The contribution is useful firstly in the customer company’s practical work, and secondly in the scientific society with the qualitative and generalized results of user studies. As the main objective of the study was to bring the end-users’ perspective to the development of the SharMe concept of users’ memory management, a wide scale of user-centered design and research methods were utilized. The study exploited research methods such as heuristic analysis, thematic interviewing and focus groups, and the main design method, Contextual Design, introduced contextual inquiries, affinity diagrams, modeling, paper prototyping etc. Contextual Design (CD) (Beyer & Holtzblatt 1998), a contemporary design method for user-centered design, is currently probably the most used even though it is rather heavy with several phases. It is an up-to-date method for executing a user-centered design process. CD method is described more precisely in Section 4.4. High user involvement during the process was assured with users acting first as informative sources in the user needs research and contextual inquiries, and then adopting a more consultative role in the paper prototype tests and acceptability evaluation tests. Although the methods resulting mostly qualitative data, with research method triangulation the results gain also slight elements of quantitative data. The tentative related applications benchmarking (see Section 6.2) and extensive user needs study produced starting point and references for the design process. With evolutionary and iterative design process the designed prototype could be constantly evaluated against the references and ideal ideas of the future system. Certain elements from Grounded theory approach (Strauss & Corbin 1990) were utilized while studying the unknown user needs and behavior. This came true by letting the users’ signals in the user needs elicitation slightly steer the research and not having too strict beforehand decided formats and expectations for the results. An ethnographic approach (Fetterman 2007) was included with highly participative field study methods, such as contextual inquiries when the research takes place in the users’ private context. The approach was present also when studying the users’ needs, group ethics and subcultures’ effect on attitudes with small groups of users.

1.3. Structure of the Thesis

So far, the introduction has revealed the background and motivation for this study. The scope of this thesis, the objectives for the research and the methodology for conducting the research have been presented. Chapter 2 shortly presents the SharMe concept which acted as a conceptual framework for this study and more or less defined the scope of this thesis.

1. Introduction 5

Chapter 3 gives an overview of the area of personal content management. Being a rather novel research field, the terminology is defined and personal content management’s relation in personal memories is discussed. The various activities around personal content are examined, and the wide spectrum of current PCM systems is shortly introduced. Chapter 4 describes the user-centered design approach in designing products. The variety of field study and evaluation methods as part of user-centered design process is described. Most focus is put on the Contextual Design method and its phases. Chapter 5 defines the terms context and context awareness, and ponders the role of context information in memories. The high potential and challenges concealed in context aware systems are discussed. Chapter 6 returns to the UCD approach and the research process of this thesis. The various phases of user data gathering, analyzing, interpretation and usage, and the used methods in them are presented in detail. Again, the focus is on the Contextual Design method and its phases. Chapter 7 presents the results of the UCD process conducted in the project. The results are divided in results of benchmark, user needs study, Contextual Design process and prototype evaluation. Moreover, consolidated results from all research phases are presented in models, affinity diagrams and design drivers. The resulted SharMe prototype being the main deliverable, most emphasis is put on it. In Chapter 8, a summary of the project is given and the overall success of the study is considered. The reliability and significance of the results are elaborated. Also the suitability of the conducted UCD process for its purpose is evaluated. Furthermore, the chapter addresses the future of the research field by stating ideas for future research.

6

2. SHARME CONCEPT

This chapter describes the SharMe Concept based on which the study was planned and directed. Briefly, SharMe is a futuristic concept developed by Nokia Research Center that combines long-term life memories, personal content management and mobility with context aware systems and persistent capturing. It emphasizes sharing and the role of community in creating value for the memories. One fundamental question lies in “what makes a memory?” By being a concept of a new way to handle life memories, the focus is set five to ten years in future. Although the research area is already populated by rather famous projects, such as MyLifeBits and Mobile Media Metadata (Gemmel et al. 2002; Davis et al. 2005), this concept was thought to fit in with its mobile device usage point of view. With the mobile devices’ versatility regarding capturing, sharing, accessing and viewing the memories, it is an attractive platform for personal content management. However, capturing and sharing memories creates in addition to technical challenges, also social and legal challenges. Therefore, one of the main goals in the project was to find scenarios and technical concepts that make sense in the eyes of most users (SharMe 2007). Moreover, the concepts would be able to make it even easier for users to create memories by immediate availability and intelligent recording algorithms (deciding when it’s appropriate to record, what is useful to keep etc.) (SharMe 2007).

Figure 2.1. The various elements in a memory savable with a mobile phone (SharMe 2007).

2. SharMe Concept 7

As the concept of memory is very vague and subjective, several interesting elements are involved in creating, sharing and viewing the memories. Figure 2.1 describes a few elements involved in the memory that can be captured with a mobile phone. The figure nicely includes both the visual and auditory information. In addition, it takes into account the various elements that have been regarded as important in memories: textual annotations and comments, location information and emotions. The mobile device is the fundamental element in the concept since it is at the center of the recording. In the concept, the captured personal media, annotations and context information are recorded in a more or less automated fashion and stored on the mobile device and back end servers. With the multitude of connectivity means (e.g. Bluetooth and WLAN), growing computing capacity and superior availability, the new generation multimedia phones are powerful tools that provide users ways to create, share and view memories all the times. These factors are essential in all aspects of SharMe and they will enable exciting community-based scenarios and concepts. The fact that phones are always carried along can be both an advantage and a disadvantage. As they are the handiest devices to record memories they are also most vulnerable to losing, stealing or just wear and tear. Hence, distributed storage solutions are considered in the concept. Regarding capturing, the concept includes ideas of making it easier for users to create memories by immediate availability of the mobile phone and intelligent recording algorithms, for example automatically deciding when it is appropriate to capture and what material is useful to keep. (SharMe 2007; Pavel 2006) Context awareness (see Chapter 5) and various metadata types allow capturing information about the user’s context to enrich the captured recording. Humans’ sensory organs are imperfect: most memories we record and retrieve later are strongly biased towards elements of our history. Past experiences and the environment we used to live in have deep impact on what and how we observe. Recording detailed memories deals with the correction of this bias: instead of partial information (e.g. a mere photo) the user will have full access to all the information the device’s sensors have access to. For creating richer memories and for making it easier to search and browse through our memories context information has to be recorded and used for annotating the media. Continuous capturing allows saving even the moments that usually are missed with traditional capturing devices. The idea of recording one’s long-term life memories is not a new one. The initial vision was developed by Vannevar Bush (Bush 1945). Notable research projects have been conducted since then in various groups. While the topic of continuous capturing still generates various controversies either due to gadgets that need to be worn or due to ethical and legal issues involved, recording the most precious, but usually missed moments, interests users (Gemmel et al. 2002; Healey & Picard 1998; HP 2007). Having certain advantages, such as allowing people to focus on experiencing the moment with little or no interaction with the capturing device in stead of

2. SharMe Concept 8

experiencing through an objective, continuous capturing has split opinions when it comes to losing privacy and control over what and when is recorded (SharMe 2007; Pavel 2006). The very same rule applies than in any capturing: users have to be in control and aware of what is recorded. Some individuals are ready to give up this control, but most of the people are not (Gemmel et al. 2004; Hayes et al. 2004). Generally in SharMe, a captured memory involves several elements, such as auditory and visual elements and various context data types (location, mood, proximity etc.). Sharing is one of the most important topics in SharMe and it introduces also various challenges. Sharing has to be made easy and safe. The user must have a control upon the memories being shared and the circle of people with whom a particular memory is shared. As a result of memory sharing new communities emerge and existing communities are strengthened. It is important for SharMe that recording and retrieving memories is not an activity that is centered on single users only. In real life, memories are often exchanged in small circles of friends and family or among a community of common interest. For this reason, the aspect of sharing memories is of utmost importance for SharMe. (SharMe 2007; Pavel 2006) SharMe concept introduces ideas of using semantic meanings in memory managing, i.e. algorithms for processing recorded data by creating higher-levels of abstractions that would allow for more user understandable representations and querying. As the memory exploration takes place on the mobile devices as well as other devices, various visualizations paradigms for various information are taken into consideration both for desktops and mobile devices. The mobile device receives again a special consideration, as it allows users to access their memories, which could be stored anywhere, at any time. In particular, the aspect of presentation of collective memories to a particular group of interest is important to consider since, most likely, different devices will be used for presentation than originally for the recording and even the single user case. (SharMe 2007) Retrieving memories plays an important role in SharMe. As stated by Gemmell et al. (2002) people have, over the history, less and less invested in actually remembering things and more and more relied on being able to find something when needed. In our accelerated world, finding information upon demand has become possible through the use of search engines. Such instant access to web content is not yet the norm for people’s personal histories. However, it would often be useful to “Google one’s life” to find past things. A search keyword would bring back the things you have seen, heard, touched, or tasted. In many cases, these keywords are actually context parameters, i.e., people remember that they were in a certain situation or heard a certain notion, story or made an association with something (SharMe 2007).

9

3. PERSONAL CONTENT MANAGEMENT (PCM)

This chapter describes what personal content management (PCM) is and what kind of services there are currently for managing and experiencing personal content. Before describing personal content management the concept of memory and its role in content management has to be understood.

3.1. Defining Personal Content Management

Personal content management is a rather fresh term to describe users’ interaction with their digital media content, such as digital photos, videos, audio and documents. Here, management involves the entire interaction process around the content: capturing, storing, processing, organizing, retrieving, sharing and deleting. Personal content distinguishes the content of the service provider and the users in the services. Related terms user-generated content and user-created content are broadly used to describe only the content the users have created themselves, while personal content is regarded as any personal files that the user has in control – also the commercial content and content created of provided by others. The above definitions vary a lot in literature and content management applications, and little differences can be pointed out. Aaltonen (2005) defines personal content as “Data targeted at human access, including individual data objects and combinations and collections thereof. It is meaningful to the person dealing with it. In addition, the person has a voluntary relationship with personal content and is in possession of it and has control over it.” This definition clearly describes content as something totally under the user’s control and meaning something for the user. According to Aaltonen, the content is defined by the user: content of one user may not be regarded as content of other users. This is in line with Lehikoinen et al. (2007) who state that data is content if the primary user is a human – not a machine. However, the definition by Aaltonen does not include the content that is not constantly in possession or control of the user but could still be regarded as content (e.g. photos stored in external locations or content that is accessible for other users’ annotations or other editing actions, too). In addition to the most traditional content types, such as photos, videos, music, messages, and chat logs, Lehikoinen et al. (2007) give examples of also other interesting – and not that obvious – content types:

3. Personal Content Management (PCM) 10

o For playing: saved games, game characters, high scores o For expressing personality: skins, profiles, presence information o For analyzing: personal health information o For navigation: bookmarks, maps, landmarks

(Lehikoinen et al. 2007) Considering the definition of content, it is obvious that these are meaningful to the user and in user’s possession. This list brings out also the aspect of content being a link to other content and still being considered as content itself (e.g. bookmarks). Also, the personalizing tools, such as skins and profiles are without doubt content although the usage of them is not always evident. In personal content management the memories are mostly related to certain events. Memories can contain photos, videos, audio, textual information etc. but the actual user’s memory of the situation contains emotions, smells, voices etc. Hence, the digital memories act as triggers for the user’s own memories (Croft 2007). When talking of content and memories, the difference between terms have to be delineated. In this thesis, with content is meant the mostly digital content that the user uses and interacts with. The interaction can be for example, sharing, browsing, capturing or editing. The content can be, for instance, photos, videos, albums, texts, ring tones or digital news. The first ones in this list refer to user-generated content (UGC) that have been created by an individual without having commercial purposes (Kiljander 2004), while the latter refer to commercial content. Here, the form of the content is not that important as the fact that the content is something interacted with (Aaltonen 2007). In addition, recordings represent the self-created content usually captured with a camera or an audio recorder. Content items refer to single photos, texts or other single content objects. Also, with media files is meant the content items but mostly the visual or audio data. With a memory is meant a slightly broader entity than content. It can be a set of photos or a single content item. The point is that the memory matters a lot – more than basic content – to the user and it is valuable and sensitive. Thus, memory can be seen as a content that has gained extra significance by, for example, annotations, metadata, being a nice take from the real world or by illustrating something important to oneself. Content can be initially regarded as memory and it is published for memory sharing purposes, or the content becomes important enough only after publishing it, and then becomes regarded as memory. Nevertheless, in this Section is presented also the human memory system, which should not be confused with the concept of memory. Moreover, digital memories are used to stand for memories in certain sections to diminish the risk of confusing the human memory system and digital memories.


3.2. The Nature of Memory

Human’s memory serves as the recorder of past, present and future. By using the memory human is able to remember events and things in past and use that information in current and future actions. The function of the memory is based on human’s perceptions, experiences and thoughts. The physiological basis for memories is the neural system in human brain. (Sinkkonen et al.2006)

3.2.1. Operational Principles of the Human Memory The operation of the human memory is divided into three levels: sensory memory, working memory and long-term memory. The sensory memory acts as the recipient of sensory information (vision, hearing etc.), and thus is very short. The working memory acts as an integrator of both sensor information and the existing long-term memory information. Only the information in the target of attention is transferred to working memory. In addition to the central executive, the working memory consists of the phonological loop and visuo-spatial sketchpad, which store the sound information, and the visual and spatial information. The new sensory information is processed taking the existing memories into account. Thus, the existing memories affect on perceiving current environment. The long-term memory stores part of the working memory’s information in the brain’s nerve cells. Hence, it contains all the past information that human needs: learnt material, operations models and rules etc. (Sinkkonen et al. 2006) The long-term memory is divided into declarative memory and procedural memory (Anderson, 1976). The declarative memory can furthermore be divided into episodic and semantic memory that store facts, things and events (Anderson, 1976). The procedural memory contains skills, such as driving a bicycle or swimming, which can not necessarily be described declaratively, but are self-evident for a human. With such division it is important to realize that the various parts of the memory constantly co-operate and all of them are required for memory to function properly. This division is more like an abstract model to express the versatility of the human memory.

3.2.2. The Concept of Memory The most interesting aspect in content managing is the concept of memory. Here, the concept refers to the factors and components of a single memory, such as a memory of last Christmas or an old friend. The memory can consist of various factors: people, places, feelings, environment etc. There is a variety of memory types. A memory of an event comprises various aspects related to a certain event. For example, a birthday party or a meeting with a friend. A memory of an object involves various kind of information about it. For example, a person, building, toy or any other object with things and events related to it. A memory of an action (e.g. memories of jogging, closeness, giving a presentation) is a compilation of several real-life events with a similar situation.


Generally, people remember best the personally significant things and those that are easily attachable to previous information in the memory (Sinkkonen et al. 2006). The length of the memory can be very varying: from a short span of time where something interesting or outstanding occurs (e.g. a car crash) to long-term memories, such as a vacation or a memory of a certain person. Moreover, the memory of an object might be scattered over time. For example, a memory of a late relative might form of different events and emotions from several times. The memories of events are stored in the episodic memory, but the procedural memory completes the memory with facts about the environment and people. Also, the declarative memory adds value to e.g. verbal stories. (Sinkkonen et al. 2006) The memory evolves over time by gaining new aspects and loosing the unused and weakest engrams (Sinkkonen et al. 2006). Every time the memory is recalled it is interpolated with the factors of the recalling situation and environment. The less important factors in the memory that are not stimulated gradually fade away. Thus, a memory of, for example, an event might differ tremendously between attendees after a few years. Moreover, the older the memory, the wider the scope of the memory is. For example, events from decades ago are thought as one memory as newer events might include several distinct memories. The extent of a memory varies tremendously. One might think every single photo as separate memories, as others think the memory is a longer span of time with several recordings and attached metadata. This largely depends on the object of the content, too. A single landscape photo might evoke more memories than a series of sequential photos of one activity.

3.2.3. Recalling Memories Recalling the memories is based on either knowing it by heart or recognizing an object or an emotion. The recalling factor can also be called the trigger of a memory. Various aspects in the memory may act as triggers: sounds, music, sights, shapes, senses of taste, words, people, emotions, facial expressions etc. It depends greatly on person what is the most effective trigger. Moreover, the trigger and its nature might change over time. Often, a smell is a very efficient trigger for very long-term memories, while visual hints are usually the best to evoke recent memories. The triggered memory might also change over time, since most triggers are related to several memories (e.g. one’s mother may act as trigger for very various kinds of memories). An overarching factor in memories is the strong emotional charge in them (Sinkkonen et al. 2006). Whether it was positive or negative, the emotion linked with the memory is often the very reason why the memory was stored in the long-term memory in the first place.


3.3. User Activity around Personal Content

Content management has evolved tremendously from paper albums to digital capturing, editing and organizing of content. As the new web2.0 applications have started the new era of user created content in web pages, sharing the content has rocketed consequently. With the recent convergence of mobile and Personal Computer (PC) platforms, the content management has become more multifaceted also in the mobile environment. To distinguish content management from digital file management, there can be found an extensive set of actions for content management. The various activities with personal content are described next.

3.3.1. Capturing Content Studies of media capturing (Okabe 2005; Koskinen et al. 2002) and usage (Van House 2005) have shown inherent needs for capturing and sharing personal content. According to Okabe, camera phones are used for personal archiving, intimate visual co-presence and Peer-to-peer news and reporting (Okabe 2005). Personal archiving is for example visual note taking and moments of the mundane life. Intimate visual co-presence refers to reporting to intimates about own current status, especially when in an interesting situation. Peer-to-peer news and reporting is about sharing newsworthy events for others with almost a journalistic flavor. The content itself becomes then the topic of conversation. Now that the content provides diverse information to act as trigger for memories, the captured content gains value in reminiscing and sharing the memories. Capturing in Bursts Other studies have pointed out that content is usually captured in bursts (Jacucci et al. 2005). The capturing situation is connected to a real-life event, and thus it is most natural that several recordings are made in a particular span of time. “Interesting characteristics of large-scale events are their spatial distribution, their extended duration over days, and the fact that they are set apart from daily life.” (Jacucci et al. 2005). These attributes partially apply to smaller events, such as birthdays, social gatherings and cultural events, too. The overarching element is the subjective interestingness and newsworthiness of the events. The interestingness factor depends greatly on users. For example, social interaction aspect, general publicity of the event or the interestingness of the content of the event can all be enough to exceed the limit capturing memories. In addition to event-based capturing, mundane recordings are captured. They represent the users’ everyday life or short interesting moments in them, such as a funny or otherwise unusual situation. They serve not only as memories but also as reminders and ways to express oneself. For example, photos of things to buy or longing voice messages sent to one’s significant other. Their importance is not always clear to other than the capturer or the original receiver, and consequently they are not shared as much or as widely as the recordings taken in events (Jacucci et al. 2005).


The Social Element in Capturing A social element is usually present at the capturing situations. In large events “spectators and visitors gather in groups and invest resources (time, energy, money) to co-experience something extraordinary” (Jacucci et al. 2005). The memorable moments are experienced in groups in order to form a wide collective memory of the event, and enjoy the social interaction even if the event itself was not interesting. Jacucci et al. also found that pictures were taken not only about special situations, but often to create stories, illustrate everyday life in a funny way or to make art. Mäkelä et al. (2000) state that the participants of large public events are organized in groups of certain characteristics to exchange information, tips and jokes among themselves and also with strangers. In the study of Jacucci et al. captured photos had three distinct types: those related to the event and its performers; those that clearly represent emotionally loaded pictures having their meaning in the social interaction triggered by the picture; and those that represent a large spectrum of the social networks involved in the spectator experience. The sociality of the spectator experience was pointed out by the fact that 44% of the pictures involved people as their object (Jacucci et al 2005). Consequently, in various events the social environment (proximity and presence) is important to be involved in the content as it serves as memory of the event later on. Event-based capturing with its social aspects offers intriguing possibilities for personal content managing and sharing systems. The sharing aspect is further discussed in Section 3.3.3. Continuous Capturing Few if any of the prototypes of continuous capturing (see Chapter 2) that are developed usually for merely research use have led to developing commercial products. Nevertheless, these futuristic prototypes of persistent recording described in (Hayes et al. 2004; Gemmell 2002; Hodges et al. 2006) offer interesting ideas of capturing the most precious moments. As the moments that are worth capturing are easily missed because of the fastness of the situation, methods for continuous capturing and marking the interesting moments are needed. Moreover, the most interesting moments are desired to be experienced in real – not through a camera objective. The function principles in the above mentioned applications are usually buffer-based when the storage space and finding relevant material are not restricting factors, or continuous with very low capturing quality.

3.3.2. The Usage Purposes of Content Czerwinski et al. have made up illustrative examples of usage purposes for user-generated content. The examples bring out the diverseness and importance of content.


o Memory aid. Finding things (such as keys and eyeglasses); replaying learning and teaching experiences; reviewing research and travel; remembering names of people and places; and reviewing discussions and meetings.

o Share personal experience. Reliving experiences with lost or distant loved ones; improving communication between grandparents and grandchildren; and sharing everyday events with people separated by distance.

o Personal reflection and analysis. Understanding personal development; reviewing conflicts; finding situational patterns correlated to emotional states; and improving health via medical monitoring.

o Time management. Improving productivity at and away from the workplace; improving coordination among family, friends, and co-workers; and identifying relevant or proximate information, given the current context (including but not limited to location).

o Security. Using information for legal purposes (such as to resolve arguments and prove alibis); for security purposes (such as personal video recordings that might include evidence of, say, a possible terrorist in a public location). (Czerwinski et al. 2006)

An example of the first group – memory aid – is StartleCam (Healey & Picard 1998). It consists of a skin conductivity sensor, which is sampled by an analog to digital converter attached to a wearable computer. Images are captured by a permanently connected camera and stored in a buffer in memory. When the computer detects a startle response, i.e. a change in skin conductivity, the buffer of images is transmitted wirelessly over the Internet to a remote server. User may then view the recordings to trigger his own memory. As this utilizes the idea buffer storing it can be also regarded as a continuous capturing system. The list brings out several personal usage purposes, but nowadays the usage of memories tend to be more social with sharing and discussing the content. Of course, the personal aspect and usage purposes exist in mundane recordings (e.g. time management, personal reflection), but keeping this information private is not that obvious anymore. In other words, the private aspect of the memories has lost meaning. The memories, even though captured by a single person, become collective as other users utilize them and add annotations. People in small communities (e.g. a group of friends) look the usage of each others’ content rather liberally and the content get mixed. Furthermore, this collective point of view in memories is emphasized later on in this thesis as mostly the event and community based management is discussed. Figure 3.1 presents various content usage and application examples in relation to the recorder of the information. The table visualizes well the social vs. private usage of various information types.


Figure 3.1. Content usage examples about by whom the information is used and recorded (Czerwinski et al. 2006). As presented in the image personal content is present in all the extreme corners of the fourfold table and everywhere in between. Self-created content can be used for merely own personal purposes (diaries, personal memory collections) or shared to others (public photo albums, self-made biographies). Again, content captured by others may be used by oneself (teaching and aiding purposes) or by others (baby books, collective information databases such as wikipedia etc.).

3.3.3. Sharing Content Traditionally sharing memories has been storytelling, perhaps flavored with photos or videos. While storytelling is most important to describe the memories detailed for people who were not present, sharing with the ones who were present is more collaborative as people discuss and complete each others’ memories. The less there are memory artefacts (photos etc.) of the object or event, the more the memory grows with time (Frochlich et al. 2002). Nowadays, with digital content and sharing tools, sharing has become independent of time and space. Users are able to share different types of memories via Internet, and communicate around the content. With the recent mushrooming of personal content management applications and web galleries, such as Flickr (Flickr 2007) and MySpace (MySpace 2007), sharing the personal content has become one of the most significant characteristic of modern Internet usage. Consequently, Figure 3.1 could be updated by


slightly lowering all the usage examples as the information becomes more and more used by others. Furthermore, other social web2.0 services, such as Second Life and Habbo Hotel (Second Life 2007; Habbo 2007), have started a new era of social computing. Studies by Kirk showed that the most intensive photo editing is done just prior to sharing (Kirk 2006). This emphasizes the significance of sharing: users want to share only their best, at least when sharing in a public forum. Koskinen states that mobile content is mostly captured and shared for joking, showing affection and “creating art” (Koskinen et al. 2002). Studies of Van House and Czerwinski et al. consolidate this with their five-level classification: The content is shared for constructing personal and group memory, creating and maintaining relationships, self-expression, self-presentation, and for functional purposes (Van House 2005; Czerwinski et al. 2006). Constructing personal and group memory by means of mundane recordings aims at constructing individual or group memories of oneself. Labeling social events noteworthy by act of taking pictures merely for sharing them mutually, is for creating and maintaining relationships. Self-expression refers to, for example, artistic, funny or experimental photos to express the user’s unique view of the world. Self-presentation differs from the latter by keeping oneself as the object of the photo and, thus, presenting oneself to others. Functional photos are for, e.g. reminding oneself or others with a visual hint. This list clearly shows the motivations for sharing for small group in mobile context. It will be interesting to see how the motivations and behavior will differ when it becomes economically sound and feasible to share publicly with a mobile phone. According to the study by Van House the camera phone photos are shared with people who were in the picture or should have been. The recipient groups are usually rather constant and non-overlapping. Thus, when one receiver is given, others can be predicted. Sharing is reciprocal, but instant answer with another photo is not required. As in the mobile camera phone photos the interestingness of the photo descends with time, most sharing is done within 24 hours of image capture. The quality standards and requirements of a sharable image depend on the group. The technical quality is not the only aspect, but also the artistic or informative quality. (Van House 2005) Sharing is not all about unidirectional sending of content, but the receiving aspect is interesting, too. The recipient of an MMS or an e-mail appreciates the shared content. This normally leads to reciprocal messaging and commenting of the sent content. Thus, the act of sharing expands to multi-phased act of interaction. On the other hand, public sharing affects on other users’ needs to share their content also. Moreover, giving descriptive tags or annotations to other’s content is helpful for both the capturer and the entire community that views the shared content.


To recapitulate, the mobile sharing is mostly for real-time sharing purposes. The communication by means of photo matters a lot for users, as it serves as the discussion around the memory. Users’ need to express themselves through sharing content and interacting with each other is a universal phenomenon.

3.3.4. Organizing and Processing Content Currently people have to manage several different types of personal content. For example, the file systems in computers and mobile devices, the email folders, web favorites or history of web pages. In personal content the fact that user probably has seen the information before allows rich contextual cues such as time, author, thumbnails and previews to be used to search for and present information. (Dumais et al. 2003) According to studies by Kirk, the most common and, at the same time, time consuming content management activity is triaging and sorting of images. From the large amounts of content it is laborious to decide what to keep, delete, share and not share (Kirk 2006). Studies of Graham et al. and Gargi, Deng and Tretter have pointed out that users tend to organize their personal content to a time-based entity of recordings corresponding to the real-life events (Graham et al. 2002; Gargi et al. 2002). It is most natural to organize the content by time because people remember events in order of time well with the help of the episodic memory. In case of remarkable events, such as concerts or parties, people’s memories are highly based on the event. For example, when reminiscing people recall the event as an entity, not small bits of information extracted from the event’s context. Time also serves well as a search key because of its unambiguity. As organizing and processing the content can be troublesome, few users are ready to do it manually by themselves. Once organized, it would be even more frustrating to repeat the process if, for example, the content served better in some other order. However, automating this requires intelligent systems to create reasonable orders when needed. To decide what content is relevant or interesting is tedious for the user, but even harder for the device. (Czerwinski et al. 2006) Although it seems tempting to label most of the content disposable, in practice it is often impossible to predict exactly which ones might gain value or be needed in the future. This phenomenon is called post-value recall; the true recall value of information is not completely known until after the information is archived. Moreover, what is important to one person may not be important to another. What is important in one context (such as at work) may not be important in another (such as at home), and what is important today may not be in the future. (Czerwinski et al. 2006) Post processing content with various tools is laborious and requires skills in tools for image, video or audio editing. Fortunately, modern capturing devices create such high-quality recordings that little editing – other than for artistic purposes – is required.


3.3.5. Retrieving, Browsing and Searching Content The stored content serves as stimuli for user’s memory. People use the content to reminisce and relive past events and discuss them. Various aspects in the content trigger various memories and feelings. As mentioned before, the significance and triggering value of the elements of the memory vary between users. For example, the names of the people may arouse strong memories in someone as for some other the sounds acts as a fruitful trigger. People associate music with certain events and identify music with certain emotions (Frochlich 2004). Generally, pictures can be said to be the best triggers to evoke the memories. The content usage studies by Kirk point out interesting notes about browsing and searching content.

o Users more likely deal with recent images and only infrequently look in the past. o Users browse rather than search for images o Search is more useful in impersonal than personal collections o Users find pictures easily enough based on time-event named directory

(Kirk 2006) The most recent content is browsed most as the memory is still vivid and evolving in the human memory. As the content is used to refresh the memory shortly after capturing, the content consequently affects on the user’s evolving memory. Moreover, recording the events – even without viewing the recordings – affects positively on memorizing the event later on, since already the action of recording adds significance to the event. Thus, the memories between a user who has recorded plenty of content from an event and a user who has no self-created recordings from the same event differ surprisingly. The personal memory is not as accurate as the digital recordings are. Browsing in random order evokes more memories than searching for a one single content item. On the other hand, when browsing other’s content the personal memory aspect lacks, and the meaning of the content is not equivalent. Searching other’s content aims usually at finding a certain kind of item, for example an item that lacks from one’s personal collection. The chronological order is usually useful enough for both browsing and searching for certain content. The sequence of events is naturally easy to remember. While browsing, the continuous reminders about the events in certain time give hints about in which direction in time to browse. Recently, giving tags to content have become popular with the help of Web2.0 content managing applications. Here, tagging means adding textual annotations to the content. It eases searching and browsing the most interesting content by using tags as keywords. Tags may be user totally set or chosen from a predefined list, when grouping and organizing the content becomes even easier – at least for the application. Services, such as Flickr and del.icio.us (Flickr 2007; Delicious 2007) utilize tagging efficiently. Tags are written to describe, for example, location, people’s names, action in the recording,


capturing device and one-word descriptions, and to use these as keywords or organizing rules later on. In addition, in Flickr the tags can be given to other users’ photos, too. This brings totally new possibilities as other users might come up with more descriptive tags than the owner of the photos (Sarvas 2006).

3.4. Categorizing the Existing PCM Applications

SharMe (see Chapter 2) as a concept for personal content management involves a wide spectrum of novel features and concept-level ideas of both mobile and PC applications. Currently, not any application has succeeded to integrate all the features introduced in the SharMe in one application, but offer some aspects of the SharMe concept. When referring to SharMe related systems, three types of systems come up as most important to be taken into account:

o Web based photo and video galleries o Web2.0 content managing, social networking and blogging sites o Mobile content capturing and sharing applications

The first group share characteristics of useful content management tools and public sharing. The content is published in the gallery to get attention and comments. Users tend to share only their best, most artistic or interesting material here. Galleries offer light features for editing and organizing one’s content. Several examples of this application type exist (YouTube 2007; Picasa 2007; Aukea 2007). This application type has remained popular because of the simple idea. The second group extends the gallery point of view further. Rapid appearance of such applications have elicited a new era of personal content managing in Web2.0. As these applications aim at being the only application needed along with the capturing device the gamut of supported features is broad. The applications support importing, organizing, editing, searching and sharing the content, recipient group management, blogging, social interaction etc. Users create their own accounts and are able to personalize their user interface. Although sharing the content seems to be the fundamental idea also in these applications, features for organizing and post processing content attract users to store their media also as private in these Internet galleries. The versatile PC platform and modern fast Internet connections offer an efficient platform for personal content management. However, having the capturing device and managing application separate requires data transfer between the systems, which users regard inconvenient and the most troublesome phase. Various example applications exist (Myspace 2007; Lifelogger 2007; Orkut 2007). In modern mobile phones the capturing device and management applications are integrated. This enables new possibilities for e.g. real time sharing of content, context-


aware systems and communication tools, which the systems in the second group attempt to utilize. The aspect of sharing is naturally limited to sharing within a close group of people, but also features for public sharing via web galleries are supported. The applications in the field are still identifying themselves, and poor design slow down the popularity of them. A few examples of this application type are MMM2 + mobshare (Davis et al. 2005) and (Flipper 2007; Jaiku 2007; Futurice 2007). A major challenge in the area of content management systems is the automation of content annotation, indexing, and organization for efficient access, search, retrieval, and browsing applications. One of the major failings of current systems is the semantic gap which refers to the discontinuity between the simplicity of visual and other information features and the richness of semantics in users’ thoughts. Smeulders et al. define the semantic gap as the “lack of coincidence between the information that one can extract from the visual data and the interpretation that the same data has for a user in a given situation” (Smeulders et al. 2000). For example, user queries posed for content search and retrieval require semantic understanding from the system to offer versatile search methods and to provide the very information user is seeking for. The authors accentuate that bridging the semantic gap between user and system is a key issue in building effective content management systems.

3.5. Recapitulating Personal Content Management

Personal content management as a novel and very multifunctional concept is an interesting and promising application area for both mobile and PC platforms. The content can be used for various reasons: sharing experiences, memory aid, personal reflection etc. Dealing with users’ precious memories, the design of PCM applications has to take the various user requirements into account. Hence, the applications supporting all of this are challenging to be designed – especially for mobile phones. Capturing content is highly based on social events and is performed in temporally quite short bursts. To support capturing of all important moments, ways to capture continuously or automatically starting the capturing are needed. The sharing aspect is emphasized in social events what leads up to requirements for proper data transfer means and collective ways of use. Moreover, to be able to efficiently manage the content, features for easy organizing and browsing – for example with the help of context data or tags – are needed. In addition, ethical and legal issues with capturing and sharing files create further challenges. Although there exists several rather well succeeded PCM applications, the application area is still evolving.

22

4. USER-CENTERED DESIGN

This chapter describes the fundamental principles in user-centered design and a few methods used to conduct UCD processes. Especially Contextual Design method is described in detail as it is the main methodology used in the study.

4.1. Background

“Great product ideas come from a marriage of the detailed understanding of a customer need with the in-depth understanding of technology.” (Beyer & Holtzblatt 1998). User-centered design can be described as an overall development process where users are taken into account throughout the process and already from the beginning of the design process. UCD includes the active involvement of users for a clear understanding of user and task requirements, iterative design and evaluation, and a multi-disciplinary approach (Vredenburg et al. 2002). A company positioning itself as a service leader cannot afford not to innovate. To design products and services that satisfy users and provide an efficient way to perform a task requires thorough understanding of the users, their goals, abilities and knowledge, and the context of use. In order to achieve this goal, UCD approach differs from the previous methods, for example system-centered design, with its contrary direction of development. While former methods worked out from the engineering of the system to eventually arrive at the end-user, the UCD approach starts from users' activity and comes to system engineering last (Redmond- Pyle & Moore 1995). UCD involves users in every phase of the design process in order to improve the general usability and the system as an entity. UCD aims at trying to remove unusable design solutions and drawbacks that can affect the human-computer interaction as soon as possible. Understanding users’ requirements, expectations and motives makes it possible to design the product exactly for the right needs and usage purposes. The context of usage affects substantially on how the user’s actions are performed to reach the goals, and how the product supports that. As it takes time to learn to use new services and applications, users can only adopt new features in small steps. According to Vredenburg et al. (2002), the most common results of UCD are: external customer satisfaction, enhanced ease of use, positive impact on sales and reduced help-desk calls. This reveals not only the HCI and usability related issues but also their indirect implications related to business aspects. The user data helps to both closely

4. User-Centered Design 23

design suitable user interfaces and tentatively design new service ideas to meet the users’ requirements. The earlier the users and their requirements are taken into account in designing new products and services, the better the economical results. However, in a UCD process, fast results and use of resources has to be balanced against the reliability and width of the results.

4.2. Defining User-Centered Design

User-Centered Design may be defined in various ways depending on the usage purpose and application area of the designed product. The overarching element is the participation of the user in the design process of a product. ISO 13407 Definition for UCD ISO 13407 is a standard to describe how to conduct a user-centered design process (ISO-13407 1999). Being based on the consensus of a wide international board of researchers and practitioners of the field gives it high credibility. Rather than describing different usability methods, it describes usability at a more abstract level of principles, planning and activities. Thus, it is more or less directed at those managing the processes. An additional important aspect is that it explicitly uses the standard definition of usability from ISO 9241-11 as a reference for usability (see Section 4.2.2).

Figure 4.1. ISO 13407 UCD process (ISO-13407 1999).


ISO 13407 is based on users actively participating in an iterative design process. The iterative approach ensures that the users’ needs and requirements are taken properly into account. In addition, the design should be performed by a multidisciplinary design team that makes sure that the allocation of functions between the system and user is suitable. Figure 4.1 describes the iterative UCD process. First, an extensive insight into the users, their tasks, use context and their requirements has to be acquired. The requirements for the system have to be realized in e.g. guidelines and constraints. Based on the requirements and information about the users, design solutions are built with incorporating the existing HCI knowledge. The solutions are then evaluated against the requirements and users’ tasks. Being a highly iterative process, this continues until the system meets specified functional and user requirements. Evaluating the requirements vs. current design is a continuous process that should be performed continuously during all the phases. (ISO-13407 1999; Jokela et al.2003) Being a high level model, ISO 13407 as a standard offers merely an outline for the process. The standard does no commit itself on the practices how the phases are conducted. From the multitude of usability study methods it is challenging for the inexperienced usability engineer to choose right methods for the right phase of the UCD process. Moreover, the starting point of the model is challenging since it might be hard to identify the need for User-Centered Design in the first place. ISO 9241-11 Definition for Usability ISO 9241-11 (ISO 9241-11 1993) defines usability as “The extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use.” The used terms are further defined as follows: Effectiveness: “The accuracy and completeness with which users achieve specified goals”. Efficiency: “The resources expended in relation to the accuracy and completeness with which users achieve goals.” Satisfaction: “Freedom from discomfort, and positive attitude to the use of the product.” Context of use: “Characteristics of the users, tasks and the organizational and physical environments. According to ISO 9241-11, the UCD process should describe: The users of the system and their goals in order to design it to satisfy real needs and to be suitable for the selected user group; the context of use in order to design the system to support the characteristics of the social and physical environment; the measuring instruments to measure effectiveness, efficiency and satisfaction (ISO 9241-11 1993). In order to understand and measure these, various research methods have to be used. Challenges in Applying UCD Integrating UCD process in software development processes – or any busy business process – is challenging because UCD, with heavily iterative prototyping, demands a lot


from implementing. On one hand, agile implementation methods are required, and on the other hand, the implementation team requires UCD answers in hours, not months (Saffer 2007). UCD does not always work as planned. Relying on users in all design dilemmas may result a design with too narrow focus. For example, UCD with a product that is directed for millions of various users may become an endless design path with all the contradictory opinions and various usage situations. In that case, radical generalizations of the user data have to be done which, again, is challenging and risky. Specifying the context and the most dominant elements in it might be challenging especially in mobile context. The social, physical and mental contexts are constantly changing when location or activity change. When designing a system for leisure time activities the system’s requirements for fitting all various leisure time contexts is emphasized. Moreover, the users’ requirements related to leisure time activities are very varying and even contradictory. For example, in a bus the system should take other people into consideration but when user is alone, the system should be an active companion. It is challenging to design totally new systems based on futuristic concepts with a UCD approach. The users can not define their future needs even themselves. The new design solutions are challenging to reliably evaluate without a functional prototype. Moreover, if the concept is too futuristic and not fesible with current technology, the users’ attitudes easily get slightly suspicious.

4.3. User-Centered Research Methods

This section introduces various user-centered research methods utilized in the empirical part of the thesis. As the process emphasized qualitative research, quantitative methods are left out of the scope of this thesis. ISO 13407 process model served as an ideology in designing the research process. The used user-centered design research methods depend largely on the type of information to be gathered. User data can be gathered in various phases of the design process. First, users’ needs, requirements and habits can be gathered for obtaining understanding about the users and their motives. Here, the user acts as, mostly passive, information source. This can be completed with knowledge of the environment and culture the user works in. After proper interpretation, this knowledge can be used to steer the design of new products or updating existing ones. Secondly, user data may be test results from user tests. This knowledge serves as guidelines in design iteration and in designing future products. The user takes a slightly more active role as a “consult” or opinion expresser for the design team. The users’ notices are focused on small details in the design, and users may even produce own ideas that are implemented in the design. The higher the level of participation and trust, the better the user has to be motivated in


order to commit oneself. Thirdly, slightly differing from the user test data, the user data can be information about users’ acceptance on the design. Acceptance results act not only as guidance in the design iteration phase but also as tentative estimate about how the final product may succeed in the markets. (Järvinen & Järvinen 2004; Preece et al. 2002) A profound understanding of the users’ behavior, requirements and needs may be achieved in many ways, depending on how the research team approaches the problem. First, the insight into the users, their activities and motivations is broadly obtained. The following iteration cycles with more specific research goals fill in the necessary details to achieve a full picture of the users and their requirements. Some tasks are so location – or otherwise environmentally – dependent that the study has to be conducted in the user’s use context. Moreover, in design processes various information sources, other than users, may be used, for example scientific publications, company internal knowledge, competitors, patents etc. However, in this study the focus is put merely on gathering user data.

4.3.1. User Data Gathering Methods System requirements specification describes the requirements that the system aims at meeting. First, we need to identify the requirements and discover the current problems. The requirements can be divided into technical and physical requirements, and user and organizational requirements. In the following, only the user requirements are discussed. As the user-centered design process is based on user data, the importance of this phase can not be emphasized enough. The gamut of germane information is wide. In order to gather extensive user requirements specification we need information about users themselves, their motivations, habits and values, information about the use context, and the tasks the users perform. The data is gathered with various field study methods – often described as ethnography which is originally a method from sociology and anthropology, and aims at studying users and their behavior (Järvinen & Järvinen 2004). In addition to the users’ actions and motives, field methods also study the social and physical aspects of the use environment and the structure of the organization the work is done in. Field research methods differ from usability tests in two fundamental ways: they do not require a design or a prototype to be produced before collecting user data, and they support gathering and analyzing data about users’ work as it is currently done. An interview is a researcher-steered, and usually scripted, interaction situation. Interviews act usually as information gathering method in the early phase of the UCD process. Depending on the focus of the research, an interview can take several forms from an open, unstructured conversation to a strictly structured question-answer dialogue. The amount of attendees varies from one to even ten (focus group interviews). In-depth interviews, especially unstructured, are flexible and dynamic suiting well for


exploratory charting of user experiences and needs, statements and attitudes, and the reasons behind them. If well prepared, interviewing makes it possible to ask further specifying questions and steer the conversation in various directions depending on the user’s skills or knowledge. Being able to view the subject through the user’s eyes truly deepens the researcher’s insight into the matter at hand. Interviewing works best when the topic of conversation is mutually understood, but is highly dependent on the users’ activity. (Marshall 1998; Nielsen 1994) Focus group is a widely used method for discussing certain matters with a group of people. Although with prepared questions and steering the conversation, the focus group setting is more natural than traditional one-to-one interview. The sessions can utilize various factors for stimuli, for example prototypes, use cases, user test data etc. The group requires a good chairman to steer the conversation. With several users and little preparation the focus group results are valid and the session is easy and low-cost to conduct. Focus group can be a useful method also for interpreting, especially when various stakeholders are used to construct an extensive of the matter under discussion. A walkthrough is a related method for browsing through the gathered material or a prototype with a group in order to reveal the most significant findings or find problems in the design. Brainstorming or other innovation methods may be associated to draw up further implications of the interpreted data or design new products or services. (Marshall & Rossman 1998) Observing – the closest to ethnography – the users as they work is a challenging but often rewarding method. In tasks where it is risky or impossible to interrupt the user, such as surgery, driving a vehicle or public performing, the researcher has to be the fly on the wall. The results from users’ verbal answers and actual observed actions may differ greatly. Users cannot always put their actions into words, or they may oversimplify their descriptions. Also, even essential actions may be so automated that users forget to mention about them. Moreover, when interviewed, the users will tell what they think they need but when observing their work it might reveal even better ways to do it. Depending on the observed task, observing is challenging as little, if any, further questions can be asked. However, while observing, the user may act most naturally and the researcher may notice elements in the situation that would stay unrevealed in a traditional test setup or mere interviewing. There are risks of the user not acting naturally or, at the worst, hiding material or information. This risk reduces slightly the reliability of the observations. Nevertheless, in observation it is possible to triangulate the study by examine if the users’ actions correspond with their talking. (Nielsen 1994; Preece et al. 2002) User data may also be gathered passively by self-reporting methods or diaries. The users really have to be motivated engaged with the research and properly instructed in order to get proper results, because the methods require both time and effort from the


user. The study has to be well focused and the questions well chosen because there is no or little choices to ask specifying questions during the data writing process. This kind of approach was not used in this thesis. Scenarios are stories about people and their activities. In a scenario various context elements and a plot for action are characterized: environment, agents & actors, their goals & objectives, and actions and events. This “frame story“ serves as a stimuli, and even provocation, for the participants. Often, the scenarios are used in an imaginary situation with imaginary technology in order to give the participants more immersive sensation, and free their minds from the chains of the restrictions of current technology. Scenarios are widely used because of their adaptability with various other UCD methods, such as prototyping and interviews. (Rosson & Carroll 2001) As the interviews and observation focuses on gathering qualitative data, user surveys offer a way to gather quantitative data about phenomenon and matters that require a large amount of responses in order to be useful. Inquiries seldom reveal much new and surprising knowledge, but are usually used to verify own preconceptions.

4.3.2. User Data Analysis “After gathering data, the next step is to cluster the data into meaningful chunks.” (Saffer 2007). Traditional user research methods, such as interviewing, surveys and prototype testing provide plenty of both qualitative and quantitative data. However, interpreting the meanings behind the data and finding the implications for the design requires careful and time-consuming examining. The data can be organized and classified in various ways, for example according to significance, appropriateness or interestingness. Single user notes or answers on a single question of a survey can be interpreted and integrated into new notes with a slightly different purpose, for example representing user requirement. The quantitative data, if any, can be summarized and analyzed with mathematical methods. Usually in qualitative research, the generalization is done by raising the level of abstraction – not by finding the averages or typicality. Thus, the goal is to find common rules and features within the studied group. Depending on the designed system, the abstraction level can be lift to level of user values. This kind of philosophy requires careful interpretation but the results may then be more generally competent and even more useful in later processes than the too case-specific results with low abstraction level. Kaasinen states this well: “The emphasis of the design should be on deciding what are the targeted values to be provided to the end users and other stakeholders, and then focusing on designing and evaluating those values into the service.” (Kaasinen 2005).


In addition, the analysis can bring out new questions that need to be answered. Beforehand built hypotheses may also be compared with the results. The new knowledge by interpretation helps not only in designing the product but also in aiming the product for right user group and use context. The Contextual Design method in Section 6.4 describes the process of interpretation and organizing user data used in this thesis.

4.3.3. Designing the New Product Designing the new way to work is demanding since all the user data should be taken into account. The task can be lightened by interpreting and compressing the user data into written design guidelines (Preece et al. 2002). The guidelines may be general rules for e.g. UI design or even developed into a guiding list of features to be included in the design. Moreover, depending on the level of abstraction, the guidelines usually remain rather valid outside the research, and may also be utilized in the following designs. In addition to own results, also existing standards and design guidelines should be utilized and compared to own conclusions and guidelines. As design-evaluation process is highly iterative, prototyping can be regarded as a method to serve the design, too. Prototyping is further described in Sections 4.3.4 and 4.4.6. Storyboarding and use cases serve in designing the product in low level. They describe certain use situations and users’ actions in general level, and provide requirements for what functionality the system should enable for the user. Navigation models, UML-models and other “traditional” software design methods offer tools for strict specifying of the system for implementation needs. These are, however, left outside the scope of this thesis.

4.3.4. Evaluation Methods In addition to evaluating the design, prototyping is useful also for creating other design solutions. Prototyping may be used already in the early phase of the design to evaluate the feasibility of the overall design or in the later phases to evaluate details of the UI or functionality. The user interacts with the prototype, and possibly is interviewed during the test. During the test, the user may comment the UI, bad design is found through the use tasks, and possibly new user or technical requirements are revealed. Depending on the use purpose, the prototype can be a vertical prototype, where it executes a limited amount of the functionality of the designed system, or a horizontal prototype, where it describes the system well in a general level, but offers no or little implemented functionalities. The functionality may also be realized with a wizard of Oz –way of thinking (Dahlbäck et al. 1993). Prototyping is useful for measuring both the functionality of the system and the use experience (Beyer & Holtzblatt 1998). Prototyping, especially paper prototyping, is further described in Section 4.4.6.


As IT systems, especially mobile services, are more and more involved in the personal lives of the users and the users are increasingly dependent on these services, user trust becomes important for successful applications. Thus, the contexts of use and usage situations have to be defined together with the users, and after designing they have to be analyzed regarding expected trust in the service. “User expectations should be in balance with the actual trustworthiness of the solutions, including technical reliability, accuracy, privacy protection and user possibilities to keep the service under control” (Kaasinen 2005). The think-aloud method is regarded as very practical during user tests as it offers a lot of information that would otherwise remain unseen. In think-aloud protocol the users are asked to verbalize their thoughts while performing their tasks in usability tests. The test moderator encourages the users to express all task related thoughts, such as confronted problems, motivations and plans for actions, revealing information about e.g. level of frustration, understanding and expectations with the application, and satisfaction in the application visual appearance and usability. Hence, the researcher might find, for example, obscurities, complex navigation models and bad terminology in the design that would have otherwise remained hidden. (Nielsen 1994) Heuristics is a widely used method for evaluating the design in its various phases. In heuristic analysis the researcher acts as a user and evaluates the design by using it, and with a list of heuristic rules, for example Nielsen’s check list (Nielsen 1994). The check list includes various elements that have been recognized as elements of usable system, for example consistency, error prevention and visibility of the system status. The most significant errors can be found with only few researchers. However, the method requires rather much experience and practice of the researcher to find the imperceptible usability bugs. Neither are the users involved in any way in the method. On the other hand, the most obvious errors in early design phase can be found with relatively little effort and expertise. A few items of the Nielsen’s heuristic list and how they were applied in this study are defined in the study description in Section 6.2.

4.3.5. Design Processes Summarized In the field of UCD, a few heavy design processes have been introduced. Next, a few of them is shortly introduced merely for reference for Contextual Design. Participatory Design (PD) is based on observing the users and acquainting the research team with the users’ work by master-apprentice method. PD is one of the first design processes taking users into account, and several variations have been introduced. Generally, PD aims at modeling the users’ current work rather than merely designing UIs, and utilizes prototyping to introduce the product to the users. Users have a moral right to participate in the design of the products they will use. Built around this idea, participatory design has grown into a variety of methods involving the users.


Proponents of participatory design tend to give their subjects a somewhat more active role than those supporting the ethnographic approach. Even though PD has not gained much success outside the academic world, with similar characteristic it can be said to be the predecessor for Contextual Design (see Section 4.4). (Carmel et al. 1993) Joint Application Design (JAD) was designed in 1977 at IBM based on their previous business planning methods. It is strictly divided into phases, and based on interviewing users and users’ participation in the design. JAD is mostly based on a workshop session which should involve multidisciplinary designers – and a user. However, it only covers the early phase of design projects, and JAD does not commit itself to the rest of thedesign process. The use context gets short shrift. Even with vague theoretical background it has become successful in the industry because of its efficiency-minded philosophy. Anyway, as first of kind, JAD can be easily said to be the forerunner of user-centered design. The economical advantages on using JAD have been documented to be significant. (Carmel et al. 1993) TRUMP (Trial Usability Maturity Process) is a rather simple design process based on ISO 13407, and covering the whole product development process. It is a rather well instructed process with ten phases – starting from stakeholder meeting and defining context of use, and ending at collecting feedback from the developed product. Also, a shortened version of the process can be executed with only three phases. The use context and users activities are in a big role before defining usability requirements and designing prototypes. The user is in a central role but cost-efficiency is another attribute aspired after. (Bevan & Bogomolni 2000)

4.4. Contextual Design

Contextual Design is a state-of-the-art approach to designing products directly from an understanding of how the user works. It is a high-detail level methodology that describes exactly all the phases of the UCD process from information gathering to prototyping. The goal is to gain extensive understanding on the user’s, their work and the working context, and by using these develop a more efficient system. Here, the work is a flexible concept: it can be a small task done with a mobile phone, a several hours lasting process where several users are involved, or anything in between. The method is highly based on the multi-phased interpretation of the users and their work, and thus benefits from a multidisciplinary team performing the interpretation. By explicitly defining the work and the system, the team is able to design a system directly for the studied group of users. Of course, generalization of the results is possible when a large enough set of users are studied. (Beyer & Holtzblatt 1998) CD is based on seven phases, and each phase is based on previous phases. The phases are described in Figure 4.2 and the following sections.


Figure 4.2. The Phase Division in Contextual Design (Beyer & Holtzblatt 1998). Viewing the figure only, the method does not seem iterative – as the state-of-the-art design methods should be – but the iterative philosophy is shown in the system design and prototyping phase. Nonetheless, the method’s division into phases is not based on iterative ideology. The method requires each phase to be executed well enough so that no returns to previous phases are needed. For example, in the early phases additional inquiries and data interpretation should be carried out if the results do not please the research group. This poses challenges to implementing the CD methodology. Especially, for first timers it can be hard to predict, for example, how many contextual inquiries should be conducted.

4.4.1. Contextual Inquiries The first challenge for design is to understand the users: their needs, requirements, motivations and approach to their tasks. Moreover, even the users themselves do not know what they want nor can express their needs. The design team conducts one-on-one or two-on-one field interviews with users in that context where the studied task usually takes place to discover what matters in it. Because users have dissimilar habits and the working practices vary a lot, several users have to be inquired. A contextual interviewer observes users as they work and inquires into the users’ actions as researchers unfold to understand their motivations and strategy. Through discussion, the interviewers and user, develop a shared understanding of the work. (Beyer & Holtzblatt 1998) The method does not give direct rules or instructions for inquiry session, but describes how the session should be like. Generally used methods for inquiring are the master-apprentice model, where the user teaches, usually through practical examples, the


designer how the work is done, or the child-parent model or scientist-research object model. The traditional set-up interviewer-interviewee or guest-host should be avoided to keep the working situation as natural as possible. The matters the designer does not understand, for example because of jargon, should be asked in order to attain a holistic insight into the users’ work. After each inquiry an interpretation session with a cross-functional team is carried out. It aims at creating a multitude of notes about the task, user, habits, environment, problems etc. and drawing graphical models of them. Later on, the notes act as the building material for affinity diagram. (Beyer & Holtzblatt 1998)

4.4.2. Work Modeling The models are interpreted graphical mock-ups of the user’s work and the working environment, and the problems in them. Five different models can be created for viewing the various aspects of the work. Flow model reveals the actors that influence the work by communication and coordination (e.g. the boss gives several small tasks). Sequence model describes the order and detailed steps of the work, and the problems that underlie. Culture model describes various cultures and policies that stand in the working environment and their influences to user’s work (e.g. national culture, business & company policies, office culture). Physical model shows the physical environment as it supports the work. Artifact model reveals how artifacts (e.g. tools, messages and work pieces) are used and structured in doing the work. The CD methodology defines rather specifically what each model type should include and how to describe it. Although the relevancy of models varies a lot based on the studied work, interpretation sessions of each inquiry should produce one or more of each model. (Beyer & Holtzblatt 1998)

4.4.3. User Data Consolidation Consolidation brings data from each user inquiries together to see the common patterns and structure, yet without losing individual variation. The affinity diagram, based on the user notes, shows the interpreted issues, problems, needs and preferences across all users in a wall-sized, hierarchical diagram to reveal the scope of the problem. Affinity diagram is built by organizing the notes in various entities and finding the motives, factors and problems behind the group of notes. The groups are given descriptive titles and the titles are then organized in larger entities until all notes are included and the structure of the diagram is organized well enough. (Beyer & Holtzblatt 1998) Consolidated work models integrate each different type of work model separately to reveal the common strategies and habits of the users’ work while retaining and organizing individual differences. The goal is to produce one of each model to represent the rest of individual models and problems in them. With the affinity diagram and consolidated work models, and the found problems in them, the design team gains focus for their discussion and redesign. (Beyer & Holtzblatt 1998)


4.4.4. Work Redesign Work redesign is the most creativity requiring phase. Various methods can be used to generate ideas: brainstorming, affinity walking, building use cases etc. Redesign utilizes the consolidated data to stir up conversations about how to improve the users’ work by means of technology to support the perceived work practices. The ultimate goal is to create a totally new approach to perform the work. The redesigned work practice is captured in a vision, a story how the users will do their work in the new way we and environment design. The team develops the details of the vision in storyboards, freeze-frame sketches capturing scenarios of how users will work with the new system. (Beyer & Holtzblatt 1998)

4.4.5. User Environment Design The User Environment Design visualizes the skeleton of the redesigned system showing each part of the system, their functionality and the navigation between them. However, the UED does not tie the system to any particular user interface. It is like designing the floor plan of a new building. Seeing the outline of the system in a high abstraction level makes it easier to design the system in more defined levels of architecture modules, software object models and user interface models. “Using a diagram which focuses on keeping the system coherent for the user counterbalances the other forces that would sacrifice coherence for ease of implementation or delivery.” (Beyer & Holtzblatt 1998)

4.4.6. Mockup and Testing “The designers create prototypes to find a solution, not the solution” (Saffer 2007). After long redesigning – without users – the users are involved in the process again. The sooner the problems are found, the better – and less expensive. As early testing with paper prototypes is easy to implement and is extremely flexible, it makes it possible to iterate the design continuously and make it as good as possible to support the users’ work (Beyer & Holtzblatt 1998). Rough paper prototypes of the system design, with post-its as screens and menus, test the structure of the User Environment Design, and perhaps the initial UI design ideas before implementing anything into code. After testing the basic structure of the system the further testing can be concentrated on the user interface or other more detailed areas of design. Then, the prototype can be, for example, mock-up screens and menus, user interface made with a UI designer (e.g. Visual Studio), a semi-functional prototype or a Wizard of Oz –prototype (Dahlbäck et al. 1993). Prototyping is a highly iterative method, and it should be carried out in several rounds (Beyer & Holtzblatt 1998).

4.4.7. Putting into Practice In this phase, using the UED and paper prototypes, a specification document of the resulted system is written. After that, the traditional software design process – or other design process – with specification, implementation and functional testing, may be used (Haikala & Märijärvi 2000). The CD methodology does not define how the


specification document and implementation are executed. As a conclusion, the Contextual Design method is for specifying what the product should include, how it would look like and how users would use it – not for specifying how to implement the product.

4.4.8. Contextual Design Revisited Although being a heavy and multiphased process Contextual Design can be adapted for various purposes. With its efficient phase of contextual inquiries it can be used for only gathering user data for basic research, or when designing a new product CD can be used partially. A rapid CD approach has become popular in the industry where the focus normally is on designing new products and not in performing basic research and generalizing the results. Either way, CD offers a properly instructed process for conducting user-centered design with extensive interpretation.

4.5. UCD Summarized

User-centered design is not only based on the principle of involving users in the design process. It also covers the ideas of iterative design with small steps to ensure that the users’ needs and requirements really become implemented, and the allocation of functions between the system and user is suitable. UCD also involves ideas of the design being conducted by a multidisciplinary design in order to reveal the various problems early enough. Thus, these ideologies make it possible to efficiently take advantage of the user data. UCD can be applied to all kinds of design cases where it is possible to define the users, their context of use and their actions with the system. Without doubt, UCD approach is a competitive approach in designing of high-quality, usable and successful products that suit the users’ needs.

36

5. CONTEXT AWARENESS IN PCM SYSTEMS

This chapter describes the terms context and context awareness. It deliberates which elements in context are relevant in capturing, sharing and browsing memories and discusses what elements there are in the mobile context, and how these could be utilized in mobile applications.

5.1. Defining Context Awareness

People are so used to interpersonal communication that it usually is very straightforward and even automatic. People can easily understand each others’ messages even if the message as such was incomplete and vague. We can sympathize with each other’s emotions and situations. For example, in a discussion, the participants observe the gestures and tones of voice of the speaker. Moreover, participants have a conception of the topic and background of the discussion. These facts help them to understand each other well and, thus, react in each others’ messages appropriately. Human’s capability to understand context makes it possible to create conceptions of the current situation and environment. In other words, we are able to make use of implicit situational information. People can intuitively deduce and interpret the context of the current situation and react properly. (Korkea-Aho 2000; Dey 2001) However, when interacting with a computer, this profound comprehension of the context is not reciprocal. A traditional computer can not interpret user’s gestures or read between the lines by knowing the common conceptions of the current subject (Dey 2001). The situation is almost the same when you have to guess the meaning of a short incomplete sentence taken out of its original context. Providing tools for the computer to have at least slight comprehension of the context could improve the human-computer interaction tremendously. Having information about e.g. the user’s physical context would make it possible to adapt the user interface according to it or tell the user about important matters that are invisible to the user.

5.1.1. Definition of Context Context has been defined several times by different people. Environment and situation as frequently used synonyms for Context give a hint about how broad a meaning it has and what kind of aspects are involved.

5. Context Awareness in PCM Systems 37 Schilit et al. claim that the important aspects of context are: where you are, who you are with, and what resources are nearby (Schlit et al. 1994). They define context to be the constantly changing execution environment. Their definition includes the following aspects of the environment:

o Computing environment (e.g. available resources, devices accessible for user input and display, network capacity, connectivity, and costs of computing)

o User environment (e.g. location, collection of nearby people and social situation) o Physical environment (e.g. light and noise level)

(Dey 2001; Dey et al. 1999) This concept includes elements of the context that are relevant to computing devices, too. It is important to understand that different parts of the context are useful for the user and the device. In order to serve the user the device has to understand partly same elements than the user to be able to alter the computer-human interaction properly, and partly other to be able to provide new information that is unnoticeable for the user. Dey et al. define context to be the user's physical, social, emotional or informational state (Dey et al. 1999, Dey & Abowd 1999). Pascoe defines context to be the subset of physical and conceptual states of interest to a particular entity (Pascoe 1998). Dey criticizes in their latter studies these two definitions to be too specific. They say that context is all about the whole situation relevant to an application and its set of users. There is so diverse set of situations that it can not be enumerated which aspects are important in the current situation (Dey 2001). For example, in a construction site the physical environment is so dominant that social situation of the user does not matter as much. On the other hand, in a mobile device’s use context, which is described more specific in Section 5.1.3, the social, physical and computing environments have all to be taken into account when designing the applications. Consequently, Dey does not qualify any of these definitions. In stead, he redefines context as: “any information that can be used to characterize the situation of an entity. An entity is a person, place, or object that is considered relevant to the interaction between a user and an application, including the user and applications themselves” (Dey 2001). This is clearly one of the broadest definitions of context. It describes the whole scale of information that is related to context and observes it through the eyes of interaction. Several other definitions include all or some of the following aspects of context:

o Physical context (e.g. location, weather, time, noise and light level) o Social context (e.g. surrounding people and their roles, work/home, activities of

the nearby people) o Mental context (e.g. user’s feelings, current tasks).

(Lucas 2001; Schlit et al. 1994)

5. Context Awareness in PCM Systems 38 This list simply perceives the context from user’s point of view. For that purpose it is a blanket but the device’s point of view is left aside. E.g. near-by computing resources are impossible for human to perceive, therefore providing such information is important to the user. ISO standard 13407 defines the use context to include the characteristics of the user, tasks, organization and physical environment (ISO-13407 1999). It is valuable to know user’s skills, knowledge, experience, education, capabilities and habits. The task involves aspects such as goals and methods. The social and physical environment covers legislation, attitudes, standards, organization structures, heat, humidity, noise level etc. To recapitulate, it can be said that almost any information available at the time of an interaction can be seen as context information (Korkea-Aho 2000). To sum these definitions up, it can be said that the context includes more or less the following aspects.

o Spatial information (e.g. location on a map, indoor location, orientation, speed and acceleration)

o Temporal information (e.g. time of the day, date and season of the year) o Environmental information (e.g. temperature, type of environment:

urban/rural/forest, air quality, smells, light or noise level and humidity) o Nearby resources (e.g. accessible devices and hosts, points of interest and near-

by services) o Availability of resources (e.g. battery level, display, network connection,

bandwidth and processing capability) o Identity (e.g. who is the user, identification of the device and user’s usage

history) o Social situation (e.g. user’s availability and proximity: who you are with, people

that are nearby and their influence on user’s actions) o Organizational situation (organizational and societal culture, position in the

organization, general attitudes and common goals) o Physiological measurements (e.g. blood pressure, heart beat rate, respiration

rate, muscle activity, perspiration and tone of voice) o Mental situation (e.g. moods, feelings and spirit) o Activity (e.g. talking, reading, walking, working, user’s level of attention and

free hands, device mode, task goals and hurry) o Schedules and agendas (e.g. calendar notes and to-do lists, future activity)

(Dey 2001; Korkea-Aho 2000; Lucas 2001; Schlit 1994; Wilhelm et al. 2004). As the context data types are so diverse, it is fundamental to make generalizations of what context data is important in general. Dey states that the context data should primarily answer to questions: who, what, when and where (Dey 2001). This indicates

5. Context Awareness in PCM Systems 39 that the most relevant context information would be identity, activity, time and location. Identity determines e.g. what features and information should be visible for the user. Activity tells e.g. how much attention the user can pay for the system. Time gives hint about in what kind of situation the user might be in (sleeping, summer vacation, working etc.). Location ties the situation in a certain space and may delimit the possibilities of the user’s action status. For example, at work the user is probably working and when moving the user might be driving a vehicle. Moreover, time and location are said to be the most relevant information for the user to be able to identify situations from the past. Of course, the relevancy of each type of context data depends always on the application and its usage purposes.

5.1.2. Definition of Context Awareness Briefly, by context awareness is meant the capability of a system to react or adapt to the available context data in order to ease the human-computer interaction. The idea of providing context awareness is to narrow the gap in the interaction between human and computer by providing the computer a primitive comprehension about the current context. Like the term context also context awareness has several definitions. One of the earliest definitions of context-aware applications was given by Schlit and Theimer. It extended the definition from applications that are simply informed about context to applications that adapt themselves to context (Schlit & Theimer 1994). In many definitions context awareness has become somewhat synonymous with other terms: adaptive, reactive, responsive, situated, context-sensitive and environment directed (Dey 2001; Dey & Abowd 1999). By the definition of Pascoe (1998), context-awareness is the ability of devices to detect and sense, interpret and respond to aspects of the user's local environment and the computing devices themselves. Dey defines context awareness: “A system is context-aware if it uses context to provide relevant information and/or services to the user, where relevancy depends on the user’s task.” (Dey 2001) This definition includes only the response to context, allowing the detection and interpretation to be performed by other systems. By Korkea-aho’s studies one definition is “A system is context-aware if it can extract, interpret and use context information and adapt its functionality to the current context of use.” (Korkea-Aho 2000). As a conclusion, it can be said that context awareness involves all the elements of detecting context data, interpreting it and using it to ease the user’s current task. According to Tamminen et al. (2004), Pascoe et al. (1999) and Dey & Abowd (1999), context-aware applications share some common characteristic features:

5. Context Awareness in PCM Systems 40

o Firstly, information and services can be presented to the user according to the current context. For example, information on what public services there are near-by or the user interface adapts its functionality depending on user’s location or current activity.

o Secondly, automatic execution of a service when in a certain context. For example, an action could trigger when user arrives home or the application changes its output devices’ properties (brightness, volume) according to the current physical context.

o Thirdly, tagging of context to information for later retrieval. For example, a camera could attach location or proximity tags to a photo in order to ease the later retrieval of that specific photo.

Properly implemented, context awareness has extensive possibilities to improve the human-computer interaction. Providing only essential information in situations where the interaction has to be minimized is a highly respected feature but requires careful design. Automatic event-triggered functions and UI that suggests actions for the user, such as Pandora (Pandora 2007) and Amazon (Amazon 2007) enable richer user experience in context where the user is actively browsing or searching information. The possibilities seem to be unlimited, but extensive research in UI design has to be conducted to be able to utilize them properly (Kaasinen 2005).

5.1.3. Mobile Context In the mobile environment a few special elements can be highlighted both from the user’s and system’s point of view. Mobile context is much more dynamic than, for example, home context. The fact that context and its properties are highly dependent on location explains why mobility adds so much variation in the context. Secondly, the mobile device’s limited battery and data transfer resources, and harder user interface restrict its computational effectiveness. The mobile context is constantly changing. It changes as the user’s physical or social environment, activity or moods change. The user needs for interaction and task execution change rapidly. A very small change in the context, such as moving from outdoors to indoors, or change in user’s activity can require mobile system’s adaptation to the new conditions. The mobile device should thus be able to adapt itself to plethora of various usage situations. Compared to a PC, whose use context is rather constant, the mobile device should be on hand in situations such as exercising, on the road and in different working conditions. Still, it should be usable in constant home environment, too. (Kaasinen 2005) A few-year-old theory describes three restrictive aspects in the mobile context (Väänänen-Vainio-Mattila & Ruuska 2000). The technical infrastructure is a limiting resource and does not cover all areas. Low bandwidth and disconnectivity bothers users

5. Context Awareness in PCM Systems 41 even today. Sophisticated applications require 3G services and high data transfer rates, but the limitations delimit also accessibility of the service. As mentioned earlier, the physical context changes frequently. The mobile systems should be usable in all conditions regardless the level of light, humidity or temperature. The user’s social context may change within seconds, and so does the bilateral need for privacy vs. need for cooperative use of the mobile system. These three aspects have however been improved in the latest years. The technological restrictions are not that dominant anymore. Current challenges reside in the mobile context’s usability and ethical challenges. Tamminen et al. present user’s multitasking to be one of the major challenges in mobile system design. It means that the user’s activity is the determining point for how much attention the user is able to pay to the system (Tamminen et al. 2004). Slightest changes in the user’s environment might attract user’s attention totally. Secondly, the use context rarely enables full focus to the mobile device from the user. Considering the facts that mobile UI’s data input and output properties are limited and the overall usability is weaker than in PC systems they require a lot of user’s attention. A context aware mobile system could relieve this by adapting its behavior according to user’s activity: minimize the need for interaction when busy and offer extra information when using the system actively. The dynamic mobile context is challenging for designing context aware systems. On the other hand, exactly the dynamic nature of mobile context is the main reason why context awareness is needed in mobile systems.

5.2. Utilizing Context Awareness

The idea of services and information being obtainable according to the current location or activity unquestionably sounds intriguing. For example, at a bus one could obtain additional information about the journey (e.g. time of arrival, current speed, information about the vehicle), be able to participate in interactive games between the passengers, or check the connections from the destination. Context aware systems may help the user in various ways. Based on the system’s functional principles they can be divided into two partly opposite groups: push and pull (Michahelles & Samulowitz 2002; Häkkilä 2006). Pulling is user-initiated inquiring of information from the system. This is the traditional way for HCI, and can be seen in most modern UIs. For example, asking for information of the current network resource environment. Pushing is event-triggered and usually automatically initiated by the system. For example, an alert sound is given when battery level is too low. The two ideologies complete each other, and they can be both seen in action in the same system. For example, modern car navigators tend to be very self-initiative when telling the

5. Context Awareness in PCM Systems 42 route, but still, all the information available is not pushed to the user without user asking for it. In mobile context pushing the information is desired in situations where the information is needed instantly or when user wants to know, for example, if a friend is near-by. Pulling information should still be the primary way to be informed of less important matters. In any case, the user must have the power to decide when any information is shown.

5.2.1. Utilizing Location Awareness As the research on contextual awareness has risen rapidly during the last 5 years developing certain product groups have gained momentum in phase with the research (Häkkilä 2006). Location awareness has led the way with navigation systems, tour guides, location-aware reminders etc. The boost in location-aware systems is partly due to the existing high fidelity location technology (GPS) and the fact that location information interest users immensely. It is interesting variable enough even to be the only provided context information in a context aware system. Various kinds of context-aware systems can be developed along with it, for example presence sensors, near-by services detectors and shopping assistants (Häkkilä 2006). Current mobile systems utilize other than location data very little. Little or no data is gathered from the environment, even though it offers important information for the user. For example, in a new urban environment it would be worthwhile to know about the surrounding services. Useful information about the user is available but not utilized. For example, information about phone’s mode (silent, loud etc.), recent use and messaging history and calendar provides possibilities to guess the user’s current and future activity. This would increase device’s awareness about the user and the UI could adapt appropriately. In addition, the information about the user interests other users, too. It is helpful to know the status, for instance, when to try to reach someone. Simplified, the information on the device’s user interests other users, whereas information about the environment interests the user himself. As the information about the user is input by himself it is easier to utilize even today than the environmental information which requires various sensors and extra gadgets.

5.2.2. Entering a Context Incoming context is a term to describe a certain context that is entered. “The designer uses incoming context to determine why a situation is occurring and uses this to encode some action in the application.” (Dey 2001). For example, in a context-aware tour guide, a user carrying a mobile phone approaches an interesting website resulting in information relevant to the site being displayed on the computer. In this situation, the designer has encoded the understanding that when a user approaches a particular site (the incoming context), it means that the user is interested in the site (the why) and the application should display some relevant information (the action) (Dey 2001). Another example, the user enters a museum and wants information on a certain object. When the

5. Context Awareness in PCM Systems 43 user takes a photo of the object the mobile service could identify the object based on location information and image recognition and provide extra information about it. In this case, the location and the captured photo is the incoming context, user is interested in the object because of taking a photo (the why) and the mobile service provides a web-page with additional information of the object (the action). This cause – consequence relation is a simple model to understand how context aware systems could adapt to the context. What is challenging is to find adequate triggers in the incoming context to induce appropriate actions. Previous paragraphs describe well how the context information can be used to increase a system’s awareness. However, context awareness can be useful also for increasing the user’s comprehension of the current situation. Even though the device is normally more unaware of the situation, the device can provide the user information that the user would not otherwise be able to perceive. For example, a system could be able to sense near-by resources that the user is interested in. Usually plain information is sufficient, because the user can interpret it well. E.g. informing when a WLAN network is available. In some situations it is valuable if the system is able to interpret the information and act properly. E.g. a car’s heat sensors detect an elk in the dark forest, and the trip computer notifies the user if the animal is too near and in danger to collide. Either way, this kind of information has to be notified to the user in an appropriate way.

5.2.3. Using Context Data as Metadata With metadata is meant “data about data” (Berners-Lee 2000). Metadata describes information related to a content object in machine understandable form and it can be embedded in the object or stored separately (Berners-Lee 2000). For example, a digital photo contains both the actual content (image) and metadata about various issues related to that particular photo (e.g. time of capture, camera properties). In content management the development of physical phone UI has not been able to keep up with the rocketing amount of user generated content (photos, videos, blogs etc.) (Sorvari et al 2004). This has caused a usability challenge, which requires new ways for managing the content. Context data as metadata of content could be a solution. With the help of metadata, automatic annotation and tagging would ease efficient content management (e.g. searching, organizing) and support versatile single- and multi-criteria searches and example-based searches. Automatic and user set manual metadata tags are handy to be added during the necessary use phases, such as capturing or sharing. E.g. when shared the content could be attached with tags that tell who the recipients were. The Division into User-Initiated and Device-Initiated Metadata Traditionally in user generated content context information has been limited to file attachable metadata of the captured device and temporal information. EXIF-metadata (EXIF 2007) has become a familiar term for digital pocket camera users. It offers

5. Context Awareness in PCM Systems 44 extensive information about the camera’s properties and settings at the time of capture. This is descriptive or syntactic metadata: data that is external to the meaning of the document itself and has a predefined syntax (Baeza-Yates & Ribeiro-Neto 1999). Unfortunately, only little of this data is relevant to other than professional users. Information about exposure time, aperture value or camera model do not interest users equally as, for example, location or people’s or object’s name in the photo do. Of the technical data mostly only photo size, resolution and time interest the basic end user. As the Internet-based personal content management applications such as Flickr (Flickr 2007) have become general, more diverse metadata is given to media files by the users themselves. Adding metadata as tags to the content is currently very popular for offering additional information of the media content. User-initiated metadata is often more descriptive regarding the content of the media file. Furthermore, device-centered user-initiated metadata is, too, more interesting to other users than that captured by the device itself. For example, information about the camera and the photographing set-up that was used to capture a nice photo is generally speaking more useful than the technical details of the photo. But then, adding such descriptive tags to files is troublesome. Examining the tagging culture in different web-applications has proven that few users bother define several tags file by file by themselves. Section 7.1 describes more precisely the use of metadata in the examined applications. Lately, context data gathering systems with location or proximity sensors have become common in mobile phones. The location is provided either by the inaccurate Cell ID tag or by using external or internal GPS module, which is able to locate the user within a 5 meter range. Proximity data can scarcely be gathered by identifying the near-by Bluetooth devices. Even though location and proximity are considered as most interesting parts of the context data, the systems are merely unintelligent data collectors without data interpreting features. Moreover, the data available is not stored in useful places, such as attached to a media file. The previous sections introduced both the user-initiated metadata (tags, descriptions etc.) and the metadata captured by the device (technical tags, time, location etc.). Usually the user-initiated metadata is input after capturing, and the device input metadata is attached at the time of capture (Sarvas et al. 2004). The device input metadata can furthermore be divided into sensor-based and recognition-based metadata (Sarvas et al. 2004). E.g. location is purely based on sensor data, and the name of the object in a photo can be decoded by image recognition. In personal content management, both user-initiated and device-initiated customs are needed. On one hand the users want to add metadata tags themselves, and control what metadata is saved. On the other hand users want to minimize their own effort and let the system do the dirty work. It is highly user-dependent what makes the memory. Sometimes it is enough to

5. Context Awareness in PCM Systems 45 see a photo to evoke the memories of an event, sometimes non visual data, such as writings, weather or current news heads, awakes hidden traces to vivify the memory. The Purpose in Adding Metadata The desired metadata depends also on the purpose of use of the recordings. Shared memories might need different metadata than those kept private. Users who receive a shared memory are interested in the objects of the photo in order to understand the meaning of the photo. For example, information about the photo capturer’s worn clothes of that time might not interest them, even though the information would be very evocative for the capturer. Anyway, adding metadata to media items is useful for the user himself and other users, too.

Figure 5.1. The content metadata from different sources. Any metadata can act as browsing or filtering rule when browsing the recordings or searching a particular item. Figure 5.1 shows the different sources of metadata and the type of the metadata. The metadata has to be somehow categorized and classified in order to use it to sort media files according to it. For example, the location data has to be categorized in a preset set of values, such as ‘home’, ‘at work’, ‘overseas’, ‘moving’ etc. This enables searching by multiple values with defined value range and browsing

5. Context Awareness in PCM Systems 46 the recordings by any metadata type. Thus, searching and browsing of even large sets of files is made a lot more efficient. The idea of searching content based on an example media file becomes feasible through decent metadata. The system could thus use the metadata to compare different files, and find the most corresponding ones. The idea of a perfect context-aware system in the field of personal content experience includes more sophisticated features of context adaptable UI and automatically initiated context-dependent functions. Talking of personal content, syntactic metadata that describes semantics of the object (e.g. who is in the photo, what happens in the video clip) should be a de facto standard (Baeza-Yates & Ribeiro-Neto 1999). Automatic capturing by some trigger in the context or browsing features that suggest actions would enrich the personal content experience. Context aware mobile systems would offer a totally new area of added value services: information on near-by services or friends in the surroundings, suggestive features for killing time, niche marketing, novel searching services etc. Mobile context awareness can easily be said to be the future trend in mobile services. Context awareness has the power to increase the experienced value from user generated media contents by adding tags about the context and offering enhanced features for content management. As the user generated content and its sharing shows no signs of fading, one may easily claim that novel metadata tagging features are needed.

5.2.4. The Challenges of Context Awareness One of the primary challenges for context aware systems lies in the complexity of capturing, representing and processing the contextual data (Korkea-Aho 2000). Firstly, suitable sensors and programs are required to capture the information of the context. Capturing certain data (e.g. location, near-by recourses and device information) is nowadays rather easy, and the required sensors can be integrated in even mobile devices. The most challenging to capture is user-related data, such as physiological signals, mental situation and activity. Physiological signals require wearable sensors, and mental situation is even harder to find out even with high-technology sensors. User’s activity is possible to be at least partially guessed by other context information. E.g. interpreting user’s speed, mode of the device and near-by devices may give a slight hint whether the user is running or face-to-face chatting with a friend. However, this requires very intelligent algorithms to combine information appropriately and deducing the connections. Secondly, a common and explicit representation format for context information is required to enable its transfer and usage for various systems. Representing and processing data requires standardized data formats. For example, location can be expressed in different ways using different reference frames. It can be expressed, e.g. as absolute spatial location, descriptive location, or relative location (Korkea-aho 2001). If

5. Context Awareness in PCM Systems 47 information is gathered from the environment, various 3rd party services are involved as information providers. The limit values in data that trigger actions have to be defined exactly in order to enable system’s decent activity when adapting to the context. Moreover, the limiting values might be dependent on the user or the environment. Thirdly, in order to use the context information the system must have intelligence to process it and deduce the meaning. The latest issue is most probably the most challenging one because context is comprised of many various elements and making the right conclusions of them needs usually integration of several pieces of information. For example, the system could deduce that it is dinner time only if 1) The whole family is at home 2) It is late afternoon 3) There has been recent activity in the kitchen, and 4) Users’ biometric signals prove that the blood sugar levels are relatively low etc. Unfortunately, usually there are not enough such preconditions available to be able to deduce the situation exactly enough. The ideal situation would be that the system understands the same semantic meanings behind the data (Lehikoinen et al. 2007). For example, the system could understand that the exact GPS location data that is measured in the user’s home indeed means that the location is the user’s home. On the other hand, the same location in some other user’s point of view is not home. A lot of semantic data can be tied to a certain user or group of users. Understanding the semantic meanings of the data opens totally new possibilities to create truly adaptable systems. This, however, is not plausible within the next few decades. The Context in a Dynamic World Another major challenge for the context-aware systems is that the context is continuously changing (Kaasinen 2005). Firstly, the context of a certain user is changing and, secondly, the contexts between different users tend to change even more. This creates inevitable challenges for the user interface to adapt itself to the different circumstances of different users and different use contexts. The constantly changing context requires learning systems and an active role from the user in teaching the system. Personalizing the context awareness and the system’s adaptability will be most important in future systems. A challenge regarding context awareness in content management is the ethics of gathering data of other users. The ethics does not limit only photographing people but also gathering any data about them (device ID, location, last conversation etc.). It surely would be nice to know all those people in the vicinity when taking a photo. This is limited by the fact that all users do not want to get captured. Partial information about who are in the photo is only partially satisfying. Forced to manually input the rest of the names in a photo is troublesome. Methods to allow capturing yourself by only the people you trust are needed.

5. Context Awareness in PCM Systems 48 A challenge in user-initiated tagging of the content is the inconsistent annotations (Sarvas et al. 2004). Different languages enable giving a name for a tag in several forms. E.g. a photo of a lady driving a car could be given tags such as “car”, “beautiful”, “Mercedes” or “two darlings”. Retrieving the photo afterwards is could get tedious because it is hard to remember which tags were given. Accordingly, when setting metadata tags it is momentous to have preset values for them. This enables categorization of tags and thereby features to search content by certain preset tags. The above mentioned example photo could be given tags: “car” and “lady”. Moreover, retrieving becomes extra laborious if the tags are in a conjugated form, which is common in, for instance, Finnish and German. Image recognition would be most useful to relieve the task, and give the media file only predefined tags. The Usefulness of Context Information One big issue in context information is its relevancy. A human is able to automatically choose the most relevant information of the context. A computer, however, might be drowned in the vast sea of context information and not be able to choose which information is important and might require actions. The big question is how specific information of one aspect is needed to define the current context well enough? For example, how specific information the device would need about the social environment of the user? What changes in the social environment should evoke actions of the system? Maybe this is why current context aware systems are so simple. It is easier to just monitor one variable in the context and act only if the variable exceeds the trigger value. For example, monitoring on temperature and notifying when it crosses the value range where the device is functional. These kinds of systems are suitable for warning about possible harms in the environment. They do not take the whole context into account, and are thus too unsophisticated to interpret the context on user’s behalf. Furthermore, the relevancy depends a lot on the action. In one case, the location information is most relevant, and in other case the physical environment is dominant. Creating truly context aware and adaptable user interfaces requires intelligent systems to interpret the context as an entity. All, or most, aspects of the context have to be taken into account to be able to understand the meanings behind certain changes in the context. It will be increasingly important for computer systems and applications to use context information. “With an increasing diversity of computer systems integrated in our surroundings and increasing mobility of both users and hardware this will be one of the main challenges in the next years.” (Korkea-Aho 2000). Introducing context awareness as simple tagging or monitoring features makes it easier to launch more the intelligent context adapting systems in the future.

49

6. USER-CENTERED DESIGN IN SHARME PROJECT

This chapter describes the user-centered design process in the empiric part of the thesis. As a part of a project with several researchers, also other relevant study phases are shortly described to give an overview. The phases of UCD are described more or less in chronological order. The phases are described in such detail level that they could be repeated, and the results with similar user group would be similar. This chapter does not commit on the results of the different phases but they are presented in Chapter 7.

6.1. The User-Centered Design Process in SharMe

Being a novel, sensitive and highly individualistic research topic, SharMe UCD process involved users richly throughout the whole process. Altogether approximately 90 users were interviewed, observed or attended to a user test. ISO 13407 was chosen as a basis for the UCD process. First, a comprehensive conception of the various users’ needs and requirements was to be gained. This was done with two rounds of thematic interviews for various user groups, and completed with contextual inquiries. Next, by conducting contextual inquiries we were to understand the usage situations, usage sequences and models, and the factors that affect them. After that, producing design solutions and iteration by paper prototyping followed. Finally, the design solutions were to be evaluated against the user needs in paper prototype tests and pair interviews with scenario role playing. As the user needs were based on users’ current behavior and their general opinions on personal content management, users’ attitudes and values came forward inevitably. Thus, however without initial intention, our UCD process involved elements from Value-Centered Design (VCD) (Cockton 2006). Users’ values got important especially in the matters of ethical and social issues. Gaining understanding on users’ attitudes and values through contextual inquiries made it easier to interpret their motivations and behavior with personal content, and hereby design a prototype for authentic needs.

6.1.1. The Goals of the Process The overall goals in this extensive UCD process was to bring the end-users’ perspective to the development of the SharMe concept. The more specific goals could be defined in five stages:

6. User-Centered Design in SharMe Project 50

1) Recognize the user requirements and needs 2) Gain understanding on the users’ current behavior and interaction with

personal content (memories) 3) Find out how the current applications suit users’ needs 4) Design a new prototype for memory management based on the SharMe

concept and user studies 5) Evaluate the prototype against user needs

The goals remained the same throughout the whole UCD process but focus between these varied between study methods. These research questions were further defined and concretized in each phase of the study. For example in contextual inquiries, the three first of the above questions were focused on, and the inquiries’ interview questions and observation focus were based on them. However, the first UCD phases, benchmark and user needs study, focused merely on questions 1, 2 & 3 in order to not to aim the research too much at prototype design at too early stage.

6.1.2. UCD Methodology The starting point for the study was that extensive field study methods would be needed to study users in their real working contexts. First, in this context the usage of mobile phones and PCs are highly dependent on the physical and social environment. For example, camera phone photos are not naturally taken in a laboratory environment, nor are the personal photos browsed with an unfamiliar laboratory PC. Secondly, we did not have good enough preconception of the to-be-studied phenomenon that building a proper environment in a usability laboratory would neither have been feasible nor natural enough for the users. In observation, the level of definition of the user’s activities defines the scale of activities that are to be studied. It might be very narrow, such as a certain single short task in the user’s work, or very broad, such as in this case, where the observed tasks in inquiries were not exactly known beforehand and the PCM involved very diverse tasks. Our initial vision was that field study methods would help in gathering first-hand information about the target group users’ work, avoiding misinterpretations of the users’ needs and requirements, and avoiding error in the results caused by unnatural test arrangements or studying wrong users. Having a several researchers’ study group in the beginning of the study steered us to using multiple methods to gather user data, and thus take advantage of the multidisciplinary know-how we had. In such qualitative research, the reliability of results is often questioned. One way to verify the results is to use multiple methods and researchers to study the same phenomena or object. This is known as triangulation and means that the overlapping results are compared to each other to find out any inconsistencies or conflicts. Use of several overlapping study methods enabled cross checking of the results and covering

6. User-Centered Design in SharMe Project 51 wider scope in the study than could have been achieved with any single method. This triangulation led to more valid and rigorous interpretation of the findings and also results that could be generalized more. Unfortunately, because of project resources and schedules the same multidisciplinary approach narrowed down towards the end of the study. However, the author of this thesis more or less participated in all study phases, so that with the tacit knowledge from various phases the results could reliably be integrated in the affinity diagram building phase (see Section 6.4.2).

Figure 6.1. The UCD process and user involvement during our research. Figure 6.1 presents our UCD process in the timeline and in which parts the users were involved. The width of each rectangle representing a phase tells broadly how wide scope it had. For example, the user needs research had wider scope than contextual inquiries. Although rather brief phases in the figure, the interpretations, especially during prototyping, were more longitudinal and comprehensive than only short steps of interpretation sessions. User data gathering was concentrated on the first phases of the process. After creating the prototype outline and starting the paper prototype tests, a more iterative approach

6. User-Centered Design in SharMe Project 52 was taken. A four-step cyclic model of “Design→ Build→ Evaluate→ Analyze→ Redesign” would illustrate this iterative approach well (see Figure 6.2).

Figure 6.2. The cyclic iterative model in prototyping. Although the iterations were not always major, several rounds of this model were executed. Because of the rather experimental UI design in this project, this approach of frequent small changes and rapid user testing proved to be most suitable.

6.1.3. Focusing the Prototype The to-be-designed prototype was not aimed at any particular user group until the contextual inquiries. At that stage, user needs study had shown signs that the prototype should be aimed at youngish technology-minded users who have at least some experience on capturing and sharing memories. Some user groups showed negative attitude towards certain elements in the SharMe concept and had strong suspicions of it. Accordingly, the prototype to be designed was not aimed at users with no experience on mobile phones and content management, or too strong suspicion of, for example, continuous capturing or sharing memories. This user group was chosen as the target group for the design phase, because they were assumed to be able to give insightful comments on the novel ideas of SharMe. They were also assumed to be the first real consumer groups of such service. In the user group division by Rogers, this group represents the first two groups: innovators and early adopters (Rogers 2003). Equally to most new products, having rather much novel features and requiring high technology, the prototype was thought to suit best this user group’s needs. This user group selection was followed throughout the rest of the study. As the bounds in this group are rather vague, little overarching factors can be found, and thus we still were able to use rather wide set of users in following study phases.


6.2. Benchmarking Personal Content Management Applications

As a preliminary study, eight PCM applications were inspected to form a profound overall insight into the current diversity of content management applications, and the usability of them. As the SharMe concept is rather extensive, various aspects and platforms had to be taken into account: mobile capturing and sharing applications, web galleries and web-based content management applications with extensive features for both managing the content and interacting with other users. The applications were investigated focusing on inspecting included features and services and evaluating the functionality with heuristics. Heuristic evaluation was based broadly on heuristics by Nielsen (1994). Each application was inspected for 5-10 hours, depending on the regarded relevancy of the application, and only by one researcher. Thus, no extensive heuristic analysis could be executed, but only a superficial inspection about the systems’ features, services and general usability. No points were given and neither were the systems put in superiority order because of their great mutual differences; instead the suitability as a reference application for SharMe prototype was pondered. With such short time, only the most relevant items in Nielsen’s heuristic list were focused on. In addition to usability heuristics, the possible success factors (e.g. in case of Flickr) and the applications’ suitability for its purposes were inspected shortly. Next, the heuristic rules that were focused on, are explicated.

1. Visibility of the system status The system should always provide the user information about what is going on, where the user is in the navigation and what happens if something is pressed. The significance of this was emphasized especially in the mobile UIs. Moreover, this is essential when dealing with users’ private memories. The visibility could rather easily be evaluated in every application by performing the various functions they offered. (Nielsen 1994)

2. User control and freedom The system has to support undo and redo in order to provide the user an option to correct the mistakes that inevitably occur. The users need a clearly marked "emergency exit" to leave the unwanted state. When dealing with precious memories, the option to correct mistakes is vital. As a whole, the user control in this kind of applications can not be accentuated enough. (Nielsen 1994)

3. Error prevention Related to the user control and correcting mistakes, the system should prevent problems from occurring in the first place. Either by eliminating error-prone conditions or checking for them and presenting users with a confirmation option before they commit to the action. The usability regarding error preventing and handling could also be

6. User-Centered Design in SharMe Project 54 evaluated well as plenty of errors occur especially when learning to use the new application. (Nielsen 1994)

4. Flexibility and efficiency of use The system should be usable and efficient for both novice and expert users. The system should allow users to tailor frequent actions and modify the UI according to their own habits and behavior. This means, for example, key shortcuts, modifiable menus and hiding unused functions. With little time, and being a novice user with all the benchmarked applications, the efficiency of expert user could not be properly evaluated. However, in case of Flickr and camera phones, this could be evaluated in the contextual inquiries. (Nielsen 1994)

5. Help and documentation It is necessary to provide help and documentation. “Any such information should be easy to search, focused on the user's task, list concrete steps to be carried out, and not be too large.” In small applications it is usually more efficient to provide the help real-time than in discrete help documentation. At its best, the instructions are an unnoticeable part of the UI but still guide the user well enough. Both the documentation and real-time instructions were inspected. (Nielsen 1994) Investigating the current applications’ features and finding the drawbacks in the UI design helped designing the SharMe prototype as a solid entity and the UI in detail. The results of the benchmark are presented in Section 7.1.

6.3. User Needs Research

Although out of this thesis’ scope and merely to provide an insight, here is described how the user needs study was conducted. The user needs study preceded contextual inquiries and the results steered also the designing of the inquiries. Moreover, the results of this requirements elicitation were integrated with results of other parts of our study to create a deeper insight into overall results. The aim of the user needs research was to tackle wide range of questions dealing with user values, expectations and emerging needs in relation to the overall SharMe concept and in specific to the processes of capturing, retrieving and sharing life memories. Interviews and scenario-based focus groups were chosen as the research methods to elicit insightful discussion with the potential target users. (Lehtonen & Soronen 2007) Interviews were carried out in two consecutive rounds with partly different sets of interview questions and variety of target groups. In the first interview round, qualitative thematic interview was selected as a suitable research method. Two of the nine interviews were focus group interviews and the rest were individual or couple

6. User-Centered Design in SharMe Project 55 interviews. To get an overall viewpoint of potential users’ attitudes to life memory management, participants were selected to represent three user groups: 1) Young families with babies or young children, 2) Young people familiar with on-line communities and 3) People living away from their relatives (in another country). (Lehtonen & Soronen 2007) In the second interview round the methodology included evaluation of the usage scenarios. Based on the findings from the first round, 11 textual scenarios were written to describe different types of use situations and aspects of SharMe. Second round was conducted with five focus group interviews with the following user groups: 1) the parents, 2) the elderly, 3) the travelers, 4) the enthusiasts (car tuners) and 5) the athletes. Overall, this broad selection of potential user groups gave a useful overview of different needs and expectations towards SharMe concept. (Lehtonen & Soronen 2007)

6.4. Applied Contextual Design

After the user needs study the Contextual Design (Beyer & Holtzblatt 1998) phase followed. While the user needs study focused merely on user needs and values, this phase inspected the users’ behavior, usage patterns and context of use, as well as verified the earlier results regarding user needs. The focus was set on both mobile and PC platforms because on one hand, the PC side with web applications represented the state-of-the-art in personal content management, and on the other hand, mobile platform as a focus in this project is the future platform for capturing, managing and sharing memories. The target group of research was the youngish people with already some knowledge of mobile devices and applications for personal content management, as stated earlier. The same group was used throughout the whole Contextual Design process. As Vredenburg et al. state, the academic study methods as such are normally too heavy for industrial needs (Vredenburg et al. 2002). In many cases some understanding is better than no understanding. In our study the resourced research and development time was short and especially the expert HCI resources were limited. These facts drove us thinking how to lighten the traditional Contextual Design methodology. In addition to insufficient resources, also another challenge existed – in the SharMe concept itself. As the contextual inquiries require existing work tasks to inquire in order to improve them, and having no existing sufficiently corresponding application, modifications to conducting the Contextual Design study method had to be done. Even though a heavy process method, Contextual Design was chosen to be used because of its suitability for requirements elicitation and user needs gathering. Consequently, a rapid version of CD was used (Holtzblatt 2005).

6. User-Centered Design in SharMe Project 56 The rapid version of Contextual Design offers a dynamic process for conducting UCD. It provides suggestions to what phases could be carried out with less contribution or left totally out. In early phase of planning CD we decided to focus on the first parts of the CD and modify the work redesigning part the most. Furthermore, having a limited amount of researchers available, certain concessions in all the phases had to be done. In user studies the importance of gathering proper user data can not be emphasized enough. Based on this idea, the contextual inquiries were decided to be conducted as the book tells. This would help also interpreting the data, which too is an important phase as it greatly affects on what kind of matters emerge from the user data. After all, we only skipped the User Environment Design, visioning and storyboarding – all of these in the redesign phase – and certain models of the work (physical, artefact and culture models). In this work, there were broadly 5 researchers who influenced on how the results from CIs were interpreted. We also used several methods to study the same matters in various phases of the research. This was seen valuable since the user groups, and thus use contexts and usage patters varied a lot between the users. As a conclusion for user data interpretation phase, an integrated affinity diagram was built to examine the uniformity within the data and reveal the possible conflicts in it.

6.4.1. Contextual Inquiries Contextual inquiries aimed at revealing the selected target users’ current behavior with mobile photography and sharing them with a PC application. Photography was chosen as currently corresponding work for the SharMe concept, as it covers well the current behavior of the users dealing with personal memories and sharing of them. The inquiries focused on finding out the usage patterns and sequences in the current user behavior. The specific focus was on revealing when, where, with whom, in what kind of situations and how users capture or share memories. Studying a phenomenon that takes place in very diverse contexts, the contexts of use were not that important focus area because the user tasks could be done in very diverse contexts. The wide range of functions in personal content management related to creating, browsing and sharing the memory objects was not only observed but also discussed with the study participants. This was to get complementary results for the user needs study, confirm uncertainties in them and possibly reveal hidden needs that even the users are unaware of. Because of the dissimilarity of study methods (interview vs. observing) the inquiries did bring useful and wide information regarding the user needs and requirements for future applications. Users Under Microscope Two various user groups – camera phone users and Flickr users – were studied to be able to observe the PCM phenomenon from different points of view. As there is no totally corresponding application and all the partially corresponding ones are not that popular, these two groups were regarded as the best user representatives for PCM applications’ users. In such low-resource research, also the challenges in recruiting the

6. User-Centered Design in SharMe Project 57 users became most dominant argument – at least at the practical level. It is critical to the success of a system that appropriate and representative users are involved in the development work (Kujala & Kauppinen 2004). However, the process of identifying and selecting users was not a focus of research because the practicality in recruiting was a dominating aspect – there was little options to choose which users to involve and which not. The camera phone users were recruited with public adverts in the Internet’s various forums and news groups. They were inquired to see how the phones’ camera features are used currently, what problems underlie and how current mobile applications and devices satisfy the users. Camera phone users’ were an important group to gather further user needs related what kind of features would be needed in the mobile environment. The Flickr users were recruited via the application’s private messaging feature. They were inquired to gain insight how users manage, view and share their memories in the PC platform. This expanded our focus from merely mobile usage to a more state-of-the-art usage of memories and other content. This helped us finding what features that are currently in the PC UI would be needed in the mobile UI and what not. Altogether ten contextual inquiries were carried out. Eight people of this group were male, two female, and the ages varied from 25 to 35. All users were Finnish but there was rather wide variation in the user characters. The research team inquired five camera phone users, four Flickr users, and one who used camera phone to capture and share photos instantly to Flickr by using the Meaning application. In that case, the inquiry focused on the cooperation of these various applications and platforms.

Figure 6.3. The users situated by information technological and photographing experience. (F = Flickr, C = Camera phone, M = Meaning; Dark blue = male, Light red = female)

6. User-Centered Design in SharMe Project 58 Figure 6.3 describes the users on after their technical and photographing experience. Although rather scattered in the area, a correlation between photography and technical experience can be seen. By leaning to our studies it can be said that using current photography and sharing tools also require technical expertise. On the other hand, as strongly technical field, photography might also interest people with an interest to IT also. Being able to inquire experience-wise rather diverse group of users gave us a satisfactory extensive sample from the population. The Practice The inquiries took place at the user’s normal usage environment, such as user’s home, work place or university campus area, in order to maximize the naturalness of the usage situation. For example, a few inquiries were held at users’ homes where other people, pets and other normal disturbing factors existed. Only three of the inquiries were held in laboratory environment because the users themselves wished that. They either did not like to capture photos in public places or they did not want the CI to be carried out in their own homes where they usually use Flickr. The users’ experience on capturing, managing and sharing memories varied a lot. This was mostly advantageous for the study as, on one hand, we were able to see how novice users regard the applications and what problems are encountered, and on the other hand we could observe the established usage patterns of expert users. The inquiry sessions were tape-recorded to be able to revise the sessions later on. None of the users resisted this, and we do not believe this to have affected on the study results. Nevertheless, the interpretation sessions were based solely on written notes because most of the sessions could be held right after the inquiries, and no backup from the recording were needed. The inquiry started with a background interview asking about normal social context, hobbies, experience on photography or similar, what, when and why capture and share personal content etc. After that, an observation and demonstration phase followed. The user was asked to do the normal daily routines with the system at hand while we observed the actions. In the demonstration part, the user was asked to perform certain tasks he has experience on, such as capturing, sharing, editing, browsing and organizing. The tasks, of course, were slightly different in the mobile and PC contexts. Here, the think aloud method was used and specifying questions were asked. During and after this, interviewing questions were asked. The tasks requiring little concentration and us having two researchers present made it possible to also ask task related questions during the task execution without interfering the user’s work. The questions dealt with the motivations behind sharing, capturing memories, the elements in a memory (context elements, audio, visual), the division between mobile and PC platform etc. Two researchers participated in each session. The other acted as interviewer and steered the session. The other wrote notes, drew models, observed the user’s actions and

6. User-Centered Design in SharMe Project 59 context, and asked additional questions when suitable. In the inquiries, observation and interviewing completed and overlapped each other nicely. The interview captured the users’ thoughts and observing their actions and intentions. Half-structured interviews made it possible to steer the topic of discussion according to the user’s experience, personal needs and areas of interest. When gathering user needs and studying what matters to the users even the irrelevant discussion with the users may turn out useful. As the users’ tasks were related to leisure time, performing them did not require full attention and interruptions were frequent in normal usage, too. Thus, interviewing during the user’s work was easy and natural for both parties. Moreover, the work patterns were not long and could be done in small steps and in various orders. The order of various tasks to be observed changed between inquiries as we wanted to follow the users natural sequences that tend to change between users. Users being mostly themselves in this context, the users’ role kept consistent through the whole inquiry. This improved the study’s reliability, because in many work related contexts the role of the user might change several times during the two-hours-session: for example from expert and teacher to subordinate. Here, the user was most natural because no artificial role had to be kept. Especially in Flickr users’ case, being able to see the users’ photos beforehand and observing the users’ actions diminished the risks of users being untruthful or rethinking their answers to please the interviewer. In getting most realistic and truthful results, of course, the interviewers’ influence was great. Moreover, there is the risk of the interviewers listening or observing only the facts that suit their thinking and research results, and that the researcher’s frame of reference effects on what is perceived (Järvinen & Järvinen 2004). The outline of Flickr users’ inquiry questions is presented in Appendix A (a translation from the original in Finnish). The mobile phone inquiry outline is not presented as there were rather little differences. Naturally, it focused on mobile content management, such as capturing photos, instant sharing and browsing photos and videos.

6.4.2. User Data Consolidation All the contextual inquiries resulted models of the current user behavior, comprehension about the division of tasks between mobile and PC usage, and users’ notices about the benefits and flaws of currently used PCM applications. In order to internalize and possibly generalize these, the user data had to be carefully interpreted. Especially when developing a totally new system, being able to see the hidden needs and possible niches is of great importance. The interpretation was carried out in several phases. First, the inquiry sessions were interpreted to create models and user notes. Second, based on the notes, an affinity diagram was built to show the consistencies and conflicts within the user data. Then, the affinity diagram and other research phases (e.g. user needs study, benchmarking) were

6. User-Centered Design in SharMe Project 60 consolidated in order to get a more solid and comprehensive insight into the user study results. Finally, this consolidated material served as basis for writing design drivers to steer the new prototype design. The analysis was redirected and iterated slightly during the process according to the inquiries’ results. For example, the integrated affinity diagram was not originally planned to be built. Interpretation Sessions After each inquiry an interpretation session was carried out. The aim of the sessions was to bring a cross-functional team to form a unified image of the users’ work. In this case, when the research team was rather small, the focus was not to share all the knowledge with the assisting researchers but to utilize them as objective commentators and to ask questions – even evident ones – to specify our insight. In a few sessions we managed to get four other participants, but unfortunately most interpretation sessions were carried out with only the two interviewers. Each participant was given a role: affinity notes writer, interaction or sequence model drawer, narrator (had to be present in the inquiry) and questioner. Each session produced 40-100 affinity notes for the affinity diagram. The notes represented e.g. user needs, motivations, researchers’ observations, interview answers and design ideas from the users themselves or brought out in the sessions. Only those phrases of the users that were considered especially important, interesting or that might enlighten or prove something special were transcribed word-by-word. The aim was to capture only one matter in one note, and avoid notes with “and”. Even though this raised the amount of notes, this was useful in affinity diagram building phase. To conserve the individual level information, the type (Flickr or Camera phone) and number of the CI was written in each note. Interaction models were drawn in most sessions. As many sequence models could not be drawn since most sequences were regarded irrelevant or uninteresting because the users’ sequences were mostly consequences of the systems’ sequences. Other CD models were not drawn as we wanted to focus on sequences and interaction. Moreover, the physical model in such dynamic context was not regarded to be of use. Neither was the culture model regarded useful, as no organizational culture present and elements of this model could be incorporated in interaction models. Also, the interesting parts of the artifact model could be incorporated in the sequence or interaction models by including the handled artifacts in the interactions. No specific information (what kind of content, who owns etc.) of the artifacts were regarded useful for this study. Affinity Diagram After inquiries the affinity notes were reclassified on an affinity diagram. The affinity diagram building aimed at perceiving the entireness of the CI’s results, and on the other hand, understanding the fundamental needs behind the user notes. Building the diagram

6. User-Centered Design in SharMe Project 61 started by reading the first notes aloud, interpreting the initial ideas behind them and starting to create groups of them. Starting to write topic level notes in the very beginning based on only one note was both useful and toilsome. Some of the topics nicely steered the organizing but some got altered several times during the building process. No topics were made beforehand. With this we wanted to ensure the material-centricity in the interpretation. The lower level topics were formulated in singular 1st person “I” –forms and represented usually user needs. For example, “I do not care for comments from others.” Even though carefully written, the abstraction level of the notes varied. Some notes represented only small details (usually problems) of the user work as others were in more general level, for example, user’s conclusion notes of the application used. In this phase the quantitative and qualitative information was merged. On one hand, each topic level note represented generalized thoughts of the users, and on the other hand, under each topic the amount of similar opinions could still be seen. No duplicate notes were removed to maintain also the little quantitative data we had. The affinity diagram was in great role in the following phases. It offered a united model beyond compare of the user data. It elicited the hidden needs (e.g. “I want publicity for my photos”) that could not be found as first hand in the inquiry notes. It provided a fruitful source for ideation and creative work. Even though a rapid version of Contextual Design, our preconception strengthened during the process that the first two phases – user data gathering and interpretation – should be done carefully. The diagram was got back to several times after finishing it: to review results and to serve as source of idea creation. Building the diagram required lots of space on walls and tables: totally 25 diagram parts – paper size of A0 – were built. With around 700-800 notes building the diagram took approximately 40 man-hours. Altogether 6 researchers participated to the building sessions in different phases. This was beneficial as especially in the beginning the notes still required some interpretation and several ears to do it properly. Although conducting this phase took very long to finish, the result was seen many times worth the effort. Consolidated Models Based on the models drawn in inquiries or interpretation sessions, consolidated sequence and interaction models were drawn. The models represented a superset of the group of individual models by generalizing them in a slightly higher abstraction level and including all the found problems in them. The consolidated models are presented in Appendix B. As the tasks observed in inquiries were divided into smaller phases (e.g. sharing, organizing, editing) the drawn sequence models represented only short independent phases of the usage. Hence, not all phases were drawn in all inquiries and no consolidated models could be drawn from all the phases of the user work. Be that as it may, the consolidated models represent the sequences and interactions far better than the affinity diagram.

6. User-Centered Design in SharMe Project 62 Integrated Affinity Diagram The cohesion between inquiries and other research phases in the project had to be confirmed. An integrated affinity diagram was built to show the consolidated research material, not just the contextual inquiries. This diagram integrated the results from benchmarking, literature review, user needs study, our own affinity diagram and another affinity diagram concerning Nokia LifeBlog built by NRC research associates. The diagram was founded on the topic level notes of our own diagram. Then, notes with broadly same level of abstraction from the other research phases were written. Fortunately, this was rather straightforward since good summaries had already been done of these phases. However, the notes represented very various aspects: user needs, habits or attitudes of the participants, UI details, general user behavior etc. Thus quite naturally, the scale of the diagram widened slightly.

Figure 6.4. One eight part of the integrated affinity diagram. To preserve the origin of the notes, a color and tag based notation was used. For example, the notes from our own topic levels remained green and orange. Using this let us see how the results from various research phases differed or corresponded. The building started by placing first our own notes, and by collecting similar notes near those notes we rather quickly got a new affinity diagram. Somewhat unexpectedly, the results confirmed each other quite well and no extra work was needed to understand the

6. User-Centered Design in SharMe Project 63 differences between them. The integrated affinity diagram represented a notably larger user sample and research methods than the first affinity diagram, and thus could be regarded as more reliable compilation. Even though more abstract, the diagram still was down-to-earth enough to guide and serve as source of ideas in the prototype design. Design Drivers & Ideas The resulted affinity diagram along with the interpreted models let us form design drivers to steer the prototype design. A list of both universal and case-specific design drivers was generated in affinity diagram walkthroughs and based on the consolidated models. As the drivers were to be the ultimate summary and conclusion of the user studies so far, we wanted to limit the drivers into reasonable amount and make them general enough to be used in other studies, too. Totally 17 drivers with various focus areas were written. The final design drivers are described in Section 7.4. The drivers served as an idea source when designing the prototype outline and as a check list during the design iteration of the prototype. Even though most drivers were not novel ones, they represented our study as best as possible, and thus had to seriously be taken into account. In addition to abstract and universal design drivers, also more detailed and user-specific design ideas were invented during the various interpretation sessions. The ideas were related to application features, services, new UI paradigms etc. The first design ideas came up already in inquiries when noticing problems in users’ actions or were further developed from users’ own ideas. Some ideas came up in the inquiry interpretation sessions or in affinity diagram building sessions. Even single user notes brought out new ideas. Moreover, simple affinity diagram walkthroughs and overviews brought out new ideas even after several rounds. The ideas were written on post-it notes and stuck in the affinity diagram near the note or group that raised the idea. After affinity walkthroughs the design ideas were removed from their original place in the diagram and were organized on a new paper wall. The wall was left in sight to provide a way to quickly review it. Because of the ideas’ multitude and irrelevance of some, they are not listed or described in this thesis but some of them can be seen in the prototype design.

6.4.3. Work Redesign Rather than being one discrete phase in the UCD process, designing the users’ work was a continuous process starting from inquiry interpretation sessions and ending at creation of the last paper prototype. As mentioned, ideas of new and better features and services came out already during interpretation phases and these new ideas guided us little by little towards a new model of the users’ work. Hence, no brainstorming sessions to create new prototype frames were needed. Instead, rather free form brainstorming between the author and his co-worker was done to clarify the ideas in our minds. In this phase the user data was already so experienced in our minds that the brainstorming was performed as more of like discussion. As the prototype outlines grew bigger than our

6. User-Centered Design in SharMe Project 64 recourses, most discussion was about on what we should focus on in detail. In some parts, such as capturing a photo, we had little to introduce, and existing solutions or merely textual descriptions could be used. Thus, we focused on the parts with most novelties: event-based management, browsing paradigms and continuous capturing (see Section 7.5). As the use cases and vision was rather clear in our minds, no graphical storyboards or visions were regarded necessary to be drawn. The schedule was tight and at this stage slightly late. Moreover, most ideas were either too challenging or pointless to be drawn. Most storyboards would have had parts of mobile UI in them, which would have made them difficult to both draw and view. SharMe concept itself included certain visions of future systems (e.g. context awareness and automatic capturing) so our idea was to balance between these initial, slightly whimsical, ideas, users’ needs and the current technology available. Moreover, practicality and feasibility were bore in mind: we did not want to create too different and futuristic model because it would have impeded the implementation and introduction to users. In such an intimate and sensitive area of research too long steps could not have been taken at one time. Otherwise, the acceptability of the prototype would suffer. Also, usability requirements for the new system were not separately conposed but – due to proper user data interpretation – we had an extensive insight into what the system should support and how efficiently. Moreover, too concrete requirements would have been impossible to measure without a functional prototype. The design idea wall and design drivers were used to steer the design, and they were resumed every once in a while. Interesting ideas were tried to fit in the prototype with still keeping in mind that the application should not get too vast and heavy. A great benefit from UCD was seen here: feasible new ideas came out already during the field study, and these – in addition to the SharMe concept ideas – could be evaluated on other users and researchers in as early phase as inquiries and their interpretation sessions. Contrary to Contextual Design, the users’ actual work was not redesigned but a new system to support their current actions was to be designed. This was because the users’ actions were mostly related to leisure time; The users were not eager to change their behavior in a voluntary activity. Moreover, little fixed patterns that would require redesigning could be found. After a few days’ idea iteration the prototype outline was clear and ready to be prototyped.

6.4.4. Paper Prototyping and Evaluation Based on the designed prototype outline, a paper prototype for mobile phone with the central functions was created. The first prototype was drawn based on fairly short thinking of what kind of screens and menus there would be and with what kind of contents. No UED was done. This was partly because the fact that the mobile phone prototype was rather simple in the UI with few screens and menu, and partly because

6. User-Centered Design in SharMe Project 65 UED was not seen as necessary when designing the prototype in low detail level. Paper prototype testing was carried out in two rounds, the first round with seven participants and the second round with five. Mostly new users were used but also some users from the inquiries were invited. Variation within the group of test users allowed us to concentrate on various areas in the prototype, and in certain tests focus on only narrow parts of the UI. The tests were conducted in laboratory environment, but any other environment with little disturbing factors would have suited. The context of usage in such early prototypes was not regarded as a significant factor. The first round of prototyping aimed at getting user’s notices on the general idea of the system, and few specific parts of the user interface design. As illustrated in Figure 6.5, the paper prototype was a low-detail set of post-it notes each describing a screen or a menu. Consequently, the layout was very simplified and only key elements were shown. In the tests the users interacted with the prototype by saying what they want to do and how, and the moderator performed the tasks and changes screens accordingly. Users were interviewed after each test task to reveal possible suspicions and define their opinions and needs that came up. Interviewing also let us inquire after ideas that were still left unspecified and required confirmation from the users. Think aloud method was utilized to catch the users’ thoughts and suspicions during the test. Once again, two researchers attended the test: the other as moderator while the other observed and wrote notes.

Figure 6.5. A part of the first paper prototype made with Post-its.


Figure 6.6. A part of the paper prototype with mock-up screen images. After each test the notes were interpreted. Comments on certain parts of the UI were written in a table showing all users’ comments on that matter. Totally new ideas were pointed out and the possibility to include those in the prototype was discussed within the research team. The first test round revealed the biggest flaws and produced improvements on the prototype’s user interface. As the prototyping went on the design both improved and was more specifically defined. The second round of paper prototype tests was done with much more detailed mock-up screen and menu images (see Figure 6.6). The mock-up images were introduced with example use cases. Using real life scenario-like use cases eased the users to understand the real implications of using the prototype as a tool for memory management. The first paper prototype tests produced implicit opinions about the prototype ideology. By using more detailed mock-ups, the second round of paper prototype tests could focus also on gathering explicit acceptability feedback. Acceptability testing aimed at gathering feedback for the overall concept and a few selected features: event-based sharing and continuous capturing, as these were the most novel features. We asked the test users to evaluate the usefulness of each feature while testing it. Especially buffer capturing was still iterated quite a lot, and new and interesting features were still invented. Users offered us valuable insight into what they regard as troublesome in current mobile user interfaces. These facts proved the significance of paper prototype testing in our study.

6. User-Centered Design in SharMe Project 67 During the first test round, the prototyping process was very iterative: after every test the most obvious flaws in the UI were identified and the related screens were redesigned – some even during the test. As mentioned, Figure 6.2 describes well the iterative ideology in our prototyping. However, altering any fundamental part of the prototype was changed only based on input from several users. When designing a new prototype of a futuristic concept it proved to be useful to perform paper prototype tests that allowed frequent iteration. Most of the unwanted or unclear design solutions or usability bugs were discovered at an early stage and the main functionalities and screen layout were redesigned immediately. Prototyping revealed also the conflicts and contradictions between the design group’s and users’ conceptions of both this particular system and the personal content management work in general. For example sequences of use, what would be the most used features, and which were mobile tasks and which PC. Furthermore, Nielsen’s heuristic check list provided a good benchmark and comparison method to confirm users’ opinions (Nielsen 1994). The prototype being so low detailed, it could not be evaluated against all the heuristic rules. The evaluation was more or less continuous and built-in during the design phase rather than a discrete phase with merely evaluating the prototype. Luckily, the users were easy to motivate as they felt being part and parcel of the design process. Hence in our opinion, measuring the system acceptability was both useful and the results reliable.

6.5. Revealing Further User Needs with the Help of the New Prototype

The final user-centered test round was conducted as a mix of acceptability testing and revealing new hidden needs by using the prototype. The aim was to reveal further user needs and hidden signals, to find out new perspectives on sharing, and to test how the prototype serve the real life event-based capturing and sharing. For example, what makes the memory, what elements from the context could be included, what kind of memories are shared with different people etc. The prototype was introduced at a low detail level and served as a framework for sharing in the scenario test tasks. Only few screens were shown – further UI iteration was now intentionally forgot. The research method in this test round was based on highly interactive pair interview with flavors from of scenario role playing. We used close real life pairs who had experience on capturing and sharing memories for both each other and other people. The users represented the same group of early adopters and technology innovators as before We interviewed totally 5 pairs: 2 married couples, 1 dating couple and 2 pairs of close friends. In the interviewing part the pairs nicely completed each other but at the same time acted as individual voices. We prepared a set of ready scenarios which in the beginning of the session were modified according to pair’s own use experience and real life events. The scenario was to provide a suitable situation for sharing and theme for discussion. After the first phase of interviewing, the users were separated to different

6. User-Centered Design in SharMe Project 68 rooms and they were given more information on the scenario they now would be in. The users were to share some of the pictures the research team brought with them with the other party and possibly add metadata or annotations to the photo. The users could decide themselves whether the picture represents a photo, video clip or something else. The users were given a free hand in using the prototype. They would tell how they would use the system to be able to inspect whether the prototype matched their expectations. This genuine-like interaction through photos made it possible to both interview the users and observe them in a simulated situation of capturing and sharing memories. The interview questions were very open, even vague, not to limit the answers already with the phrasings of the questions. As this test round was conducted in so final stage of the project, little interpretation was done. The results did not affect on the prototype design anyhow. However, the results raised a lot of open questions to be studied later on. On the other hand, the acceptability results were positive, and thus no massive actions for altering the prototype were required. Such simulative scenario-based method covering the whole usage process from capturing to sharing worked well for both acceptability testing and the quest for new kind of user needs.

6.6. Summary of the Conducted UCD Process

The UCD process was rather heavy with altogether almost 100 users involved. Starting with extensive user needs study and benchmarking of related PCM systems offered a fruitful basis for contextual inquiries. The inquiries focused on the usage patterns of Flickr and camera phone users. To interpret the user data, two affinity diagrams, consolidated models and design drivers were composed. The affinity diagrams revealed the users’ current problems and needs for new PCM applications. This consolidated data and the design ideas that had been coming up during the user studies served well when designing the new prototype for mobile memory capturing, sharing and managing. The prototype was designed with a highly iterative approach. Two major rounds of paper prototyping with users and several iterations confirmed the user-centeredness of the design. Hence, the users could affect the layout, outline and functionalities of the prototype. Furthermore, a test round with pair test methods and the prototype acting as a platform was conducted. This revealed the acceptability of the prototype but also brought out new perspectives for sharing memories.

69

7. RESULTS OF THE USER-CENTERED DESIGN

This chapter presents selected results of the UCD process in SharMe. Although there were many various phases and results of them, the focus of this chapter is on the results of contextual inquiries, the resulted prototype and the acceptability evaluation of it. The results of the user needs study are presented at more general level because that part of research was conducted by other members in the research group.

7.1. Related PCM Applications Benchmark

The benchmark as first study phase provided plenty of information about the usability and services of various state-of-the-art applications and ideas how to improve them. Section 6.2 describes more detailed the motivation and process of benchmarking. Next, five most interesting and momentous of the benchmarked and SharMe related applications are shortly presented. This group also represents the extent of benchmarked applications: mobile and PC applications and applications that utilize both mobile and PC UIs, and both simple gallery and web2.0 content management applications are covered. As the evaluated applications are so diverse, comparing their usability or functionalities is challenging and was not regarded useful in this case. Part of these benchmarked applications acted as reference when designing the SharMe prototype. Finally, tentative design ideas for the new prototype are listed.

7.1.1. Flickr Flickr is a website for storing, managing and sharing photos (Flickr 2007). It is a typical social web2.0 application including forums, user grouping and annotating other users’ content. Flickr seems to be used by both photography professionals and hobbyists, and less photography-oriented users. There is also a mobile version of Flickr which was evaluated briefly. Of the plethora of web-based personal content sharing sites Flickr was chosen because of its international popularity. Flickr offers many useful, and even more not that useful, services: View the most interesting (most commented, bookmarked or uploaded) photos of each day; Users’ testimonials of other users; Creating user groups of other Flickr users to allow various sharing levels; View which of the user’s photos are viewed recently by other users; Metadata tagging features are rich with system suggested tags, tagging other users’ content, drawing annotation boxes inside the photos and searching content by tags. Overall, Flickr provides a wide and sufficient set of features for photo managing and

7. Results of the User-Centered Design 70

sharing and interacting with other users. Fortunately, using these various features does not slow down the usage because the photos themselves are the most heavy regarding processing and transferring. The general usability is rather good. The layout is simple enough to allow effective use. Flickr shares a few good common web2.0 UI design solutions such as interestingness coded as size in tags, and UI personalization. Features that are not used are hidden from menus which, there again, creates contradictory reception.

Figure 7.1. The Flickr user interface layout (Flickr 2007). The status of the system is not always clear for the user, nor is the location in the wide jungle of users’ content. Basic usability rules, such as consistent terminology and functioning, have unfortunately not been followed well enough. Uploading tools are very inefficient. However, the existence of API interface has yielded 3rd party add-ins for uploading, and among others interesting map platforms to visualize where photos were taken (presuming that location information is attached to the photo). Flickr is adequate for basic users but for frequent usage, its sharing, organizing and editing features are not sufficient. Like in all PCM web applications the data transfer becomes


the overriding drag. To work sufficiently a broadband Internet connection is required. Despite the small usability faults, and considering the wide features and popularity, Flickr was thought to be one reference product when designing new prototypes.

7.1.2. Nokia Lifeblog “Lifeblog is a digital photo album tool designed with mobile phone photographers and bloggers in mind. Lifeblog’s phone and PC software automatically organizes your digital media between your mobile phone and PC so you can view, search, edit, and share your images and messages.” (LifeBlog 2007). Lifeblog represent both the second and third group of the above mentioned application types, but offer little communal features. LifeBlog utilizes the chronological view in browsing the content. The content consists of photos captured with camera phone, and sent and received messages. The mobile UI is sufficient for simple browsing. The PC UI with merely larger UI and only little extra features does not really add value to the use experience. The mobile version is simple and useful, but has not improved the common mobile UI usability bugs, such as awkward menu order, and low utilization of the number keys of the phone. Moreover, user is unaware which features and operations are chargeable. As mentioned, the PC version does not add in any extra features, and the better overall usability of the PC platform is thus not utilized. A few nice UI design details do not still improve the acceptability of the PC version. Moreover, required synchronization with the mobile phone makes it even more laborious. Lifeblog comprises all the content management phases, and moreover it could be thought as SharMe’s early predecessor. Thus, it is natural to use Lifeblog as a reference in the prototype design.

7.1.3. Context Watcher Context Watcher (Koolwaaij 2006; Context Watcher 2007) is a mobile application running on Nokia S60 platform, originally designed for research usage. It automatically records context data as photo metadata. For example, location based on GPS or Cell ID, bio-data based on external device and weather based on location-inferred remote weather service provider.

Figure 7.2. Context Watcher UI screenshots (Koolwaaij 2006; Context Watcher 2007).


Photos can be uploaded to Flickr or own blogging sites. Providing services by using 3rd party services makes the application more expandable but the whole application package, with various services, is then hard to maintain. Worst of all, user is left totally unaware about the pricing of various services. Context Watcher includes interesting and novel feature and service ideas, but the implementation is poor. With all the services and required extra devices to collect metadata makes the application too complex for the user. Figure 7.2 shows the layout of the user interface. The user interface is filled with small and major usability errors: terminology is poor, no consistency in the menus and functions whatsoever, poor help and instructions. As a research application Context Watcher is hard to be taken seriously as an commercial product, but the novel features and ideas make it an interesting concept. The UI should not be used as reference in prototype design.

7.1.4. Meaning – Merkitys Meaning (Merkitys in Finnish) is a simple mobile application for capturing photos with mobile phone camera and sharing them directly to Flickr. Meaning attaches various metadata to photos about location (based on GPS or GSM Cell info), nearby Bluetooth devices, and calendar events. The metadata types are interesting and the implementation is slightly better than in Context Watcher. Apart from few nice UI design ideas, the UI suffers from several minor usability bugs. Moreover, benchmarking the application came to an end prematurely because it suddenly ceased to function.

Figure 7.3. Meaning UI screenshot (Meaning 2007). Figure 7.3 shows the screen where user may choose the attached metadata tags before sharing to Flickr. One other useful feature is the summary screen which shows how many photos are published, not published, in queue, drafts etc. Idea of mobile sharing to public forum, such as Flickr, is fresh and intriguing, and provides lots of possibilities for users. Especially for users who are willing to share instantly after capturing without any editing, this is a serviceable application. Altogether, this compact application can partly be used as reference in prototype design.


7.1.5. Multiply Multiply – evaluated shortly – is one in the multitude of communal web2.0 application where users may share their media files, blogs, calendar and almost any other personal information. It provides lots of similar features to Flickr, but supports also other media file types than photos and emphasizes the interaction between users with, for example, dating and recruiting services.

Figure 7.4. Multiply UI screenshot (Multiply 2007). As the PC platform is more fruitful for usable user interfaces, the application excels in usability in the group of benchmarked applications. Figure 7.4 illustrates the general layout of the Multiply web page. Multiply has excellent content management and personalization features, and great instructions and guides. Providing such a versatile platform for personal content management and social interaction, Multiply easily takes lots of time. For especially the users seeking for publicity or respect, like in blogging, it might get laborious to keep other users interested in oneself. Yet, this is not characteristic for Multiply solely, but for all such applications. Be that as it may, Multiply is a great reference application when designing the SharMe PC UI.

7.1.6. Conclusions of the Benchmark The benchmark proved that the mobile UI design is still in its infancy when compared to PC UI design. Good UI design ideas in PC are hard to implement in mobile UI. Web based content management applications with communal aspects have gained popularity within Internet users – and not for nothing. They are easy to learn to use, very adaptable for various users, offer great content management features (sharing, browsing, tagging, organizing, searching etc.) and provide an interesting environment for social interaction. As the success of social media applications is highly dependent on social trends and even coincidence, usability of the application does not guarantee success, but it improves probabilities substantially. Supporting API interfaces enables the efficient usage of users’ resources – the users may design their own add-ins to offer more features for the entire user community. The most significant drawback in PC platform is


the lack of capturing devices, and thus need for tedious synchronization with capturing devices, such as digital camera or camera phone. The www-applications are directed for a wide range of users, but the current mobile applications require technological expertise to get them functioning properly. The state of the system is not told clear enough, and the user is often left thinking what is happening. Real-time instructing, walk-through demos and easy wizards would ease the mobilization of these applications, and help the users who use them rarely. When dealing with sensitive personal content, the performed critical operations have to be confirmed to prevent loss of important data. Currently the mobile capturing and data transfer technology, and user interfaces fail to fill users’ expectations. Needs for capturing the desired situation with a quality good enough and sharing the captured memory are satisfied only sufficiently. Mobile applications are mostly prototypes for testing, and hence suffer from lapsus in design. The visibility of system status can never be emphasized enough, nor can the importance of real time instructions. The terminology is at poor level. The phone keyboard could be utilized by offering more key shortcuts, and thus extend the set of features supported in the mobile UIs. Unquestionably, the mobile device with a small screen and limited input units is a challenging platform for content management applications. Therefore, the mobile applications’ features are restricted to only capturing and sharing the media – not browsing or searching. As Czerwinski et al. report, the required technology is available but it is challenging to form a desirable mobile application with enough features and sufficient usability (Czerwinski et al. 2006). The benchmarked mobile applications offer little reference to the SharMe-prototype design – except LifeBlog. Obviously, with rather poor existing applications and rapidly evolving mobile technology, there is major potential in mobile PCM field, and this is the very prospect area SharMe focuses on.

7.2. User Needs

The user needs are based on mostly the user needs interviews but also on contextual inquiries and the last evaluation test round where also new hidden needs were briefly inquired. As this part of the study is mostly other researchers’ contribution and not the focus in this thesis, the results are described briefly – only to give an image. The types of memories users are willing to save can be divided into three categories: happy moments with family or friends, life history of important events and development steps, and mundane details to facilitate daily errands. Hence, users are interested in saving data which captures moments of some significance. Moreover, the needs for saving are often interconnected with the idea of sharing the memories. (Lehtonen & Soronen 2007)


Most users are willing to share their memories with other people – depending on the contents of the memory and with whom to share. Although most users are willing to share either with a small group or publicly, some users regard the digital memories too private or uninteresting that they avoid sharing them. Generally, users are more interested in sharing happy memories while negative memories are more often considered to be private. Face-to-face sharing is emphasized even today as it makes it possible to naturally interact with others, tell stories and comment the others’ digital memories. This collectiveness in commenting others data proves us that memories are often created and modified together. As a derivative of real-time face-to-face sharing, some users are willing to share recordings in real-time even if apart – almost as a part of the communication between the sender and receivers. (Lehtonen & Soronen 2007) Users are generally troubled by trust issues of saving. Firstly, the digital memories need to be secured and free of risks to be deleted by mistake. Secondly, users want to control who may see their memories. They want to be sure about the intactness and privacy of the memories. Thirdly, they neither want to get recorded in others’ memories nor record any extra people into their own. (Lehtonen & Soronen 2007) Continuous and automatic capturing raise suspicions, too. Again, users are concerned that the privacy will be lost with people continuous capturing their environment. Similarly, if the capturing is not in users’ hands the device might record anything without users knowing it. Only few cases where automatic capturing would be useful came up but even these did not raise the popularity of it. (Lehtonen & Soronen 2007) When asked about what elements there are in the memory, only few context related metadata types came out. Users are mostly interested in the location and the people around them. Also time, event’s name and textual descriptions and annotations were regarded interesting and useful. Users stated that these hints about the context are enough to remember the situation clearly enough.

7.3. User Behavior and Habits

The results described in this section are based on consolidated interaction and sequence models and both of the affinity diagrams. Some issues were also inquired in user needs study. Next, the most relevant notices are shortly described. The consolidated sequence and interaction models are presented in Appendix B.

7.3.1. Events for Close Group of People The inquiries pointed out that users often capture photos in bursts. The ordinary everyday life is little recorded but when the users are involved in something interesting, abundant recordings are taken. Similar results of event-based organizing and sharing have been found out in other studies, too (Graham et al 2002; Gargi et al. 2002). In


addition, manual organizing is often done to put the content in order by time, event type or people. As the capturing is highly event-based, users reminisce about the memories as entities. This is particularly emphasized in long-term memories. Contextual inquiries brought out that groups with close relationships, such as family or a group of friends, share the need for creating collective events. As the group usually participates together in various events the recordings from different devices are then wished to be combined and set accessible for the whole group. Furthermore, if only few group members participate in an event the rest of the group wants to view the recordings afterwards, too. Most recordings are made keeping in mind what others might want to view from the event.

7.3.2. The Nature of Viewing On one hand, users like to see as much as possible at a time in order to choose certain content in closer viewing, and on the other hand, they want to see the single content object as detailed as possible. This behavior is natural for people as humans are very visual by nature and tend to view both the entireness and details. The Flickr users stated that it is very useful to see small thumbnails of photos before viewing the details of any photo any closer. For the users who do browsing, the path is usually more interesting than the goal itself. They do not need any search functionality as long as it is easy to browse the content in order of time, location or names. It is easier and more amusing to search for content by limiting the group of content objects smaller. Furthermore, users manifested that in mobile UI it is easier to browse rather than search by keywords. But then, some users need the search function and ways to quickly find certain content items. Keywords in a search can be, for example, location, people’s names and event’s name. Also certain attributes that describe the contents of the content item, such as color, style or objects in it, serve well as keywords. Sometimes they would search content by random keywords only to see what comes up. Quite surprisingly, both long-term and short-term memories are both browsed and searched for – little difference was noticed. Users want to store their content in one place only. Thus, the browsing and other managing becomes easier. Also suspicions of the durability of the content could be seen: some users want to have the physical storage in their own hands instead of trusting a remote server.

7.3.3. The Differences between Camera Phone Users and Flickr Users

The public sharing was regarded very differently within the Flickr users and camera phone users. For many Flickr users – at least in this take of the population – the point in using Flickr is publicity. As many of them are photographers, they want to get feedback


on their photos and that as many as possible would see their photos. This might be also because of the Flickr users in this study could be categorized as innovators and the camera phone users only as early adopters (Rogers 2003). This could be seen in the photography experience, too. The Flickr users were all very photo-oriented, experienced in photography and required a lot from the technology while the camera phone users were satisfied with less sophisticated technology. This could be seen in the way Flickr users wanting to share only good quality photos (artistic, well edited, good camera quality etc.). This was most probably because of the large and critical audience they shared with. Flickr users were generally more interested in metadata. This was maybe because they were used to the tagging feature in Flickr and used it a lot for searching photos. Also, they were more interested in novel technological innovations, and the constantly growing amount of information did not disturb them. However, in camera phones the need for context data was regarded smaller. This was probably because the captured photos served mostly as instant communication instrument rather than being precious memories. Also, with much poorer image quality the users concentrated to require better image quality before anything else. Overall, having metadata is regarded more useful in long-term memories where users benefit from any additional information to be able to remember more vivid. Generally in this study, camera phone users represented the basic end users while Flickr users were clearly more technology-oriented and required much more of the system they use.

7.3.4. Summary of the Consolidated Models Consolidated interaction models and sequence models of both Flickr users and camera phone users could be composed. All the four consolidate models are presented in Appendix B. Flickr users regarded the application clearly as a tool for sharing photos publicly. As Flickr users were rather photography-oriented, there could be seen several tools for photo capturing. Moreover, several means of communication – within and outside of Flickr – were used to inform other people of the shared photos. Discovered problems were mostly related to technical and functional limitations of the application and the fact that Flickr did not fulfill the expectations of communicating with other people. For example, photo editing and file transmitting features were not seen usable enough. Summarized, Flickr was not regarded versatile enough as a PCM application. Camera phones served as a tool for capturing, sharing, viewing and managing. However, the current technology (e.g. photo quality, data transfer) limited the usage to only using camera phone while mobile and other capturing devices or PCs were not


available. Although initially designed for communicating, camera the photo sharing and communicating tools of the camera phones were not seen powerful enough, and the usage too expensive. The benefit of being able to do most PCM actions with one device was diminished by the poor technology and usability. A common trouble was that users want to manage all their content with one system. It was inconvenient to, for instance, store photos in several locations. This requires remote storing solutions and proper synchronizing features. Overall, in the studied user group users seemed to have quite fixed behavior patterns and they are not eager to change them – especially within the camera phone users. For this, introduction of novel ideas in future systems must proceed gradually and carefully.

7.4. Design Drivers

After careful interpretation and consolidation, the integrated affinity diagram and consolidated models were done. Moreover, there were results from the benchmark and user needs study. To sum these results up, design drivers were written. The design drivers served as guidelines when developing the prototype. The goal was to formulate general enough drivers to be able to use them also outside this concept. Originally 17 drivers were written. However, here some drivers are combined and refined to better represent the user studies’ results. The resulting 11 drivers still represent our study results quite extensively. Next, the 11 refined drivers are listed, and a further description is given to explain the background of each one. The list is in thematic order as order of importance could not be created with such little quantitative user data, nor was it seen relevant.

1. Let the user control how to capture memories The user study pointed out that the various modes for capturing are needed. For example, single shot, burst of shots, night shooting, video capturing etc. The used shooting mode depends largely on the situation and what is captured. Especially, continuous and automatic capturing received contradictory reception. Even though the advantage of continuous capturing to capture the usually missed moments was understood and agreed, such features have still to be under the user’s control. Moreover, the user does not necessarily have time, for instance, for adding annotations or metadata at the capturing moment.

2. The system has to make sufficient backup and show the user clearly when and what is back-upped

On one hand, a single place for all digital memories is desired to provide easy access to all the memories but on the other hand, users fear loosing the memories if not stored in multiple locations. The users have a common concern about if their data survive and stay intact in both the Internet and their own data storages. Especially if the memories


are stored in the Internet, a constantly back-upping feature and informing the user of it are needed. In addition, because of human errors and accidental deletion, the back-ups must be easily accessed to restore the lost data. This sounds like a trivial design challenge but the users’ suspicions are not too easily neutralized.

3. Provide an option to edit before sharing Although seen as troublesome phase, editing is wished to perform before sharing the recording. As the camera phone is not usable platform for editing recordings, user not wanting to share before editing creates challenges for mobile sharing. The advantage of instant sharing is impaired if toilsome actions are required first. This requires high quality of the mobile phone camera and other capturing devices, and perhaps features for automatic post processing.

4. Ease the task of adding metadata Metadata is regarded as useful but the manual adding – especially with a mobile phone – is regarded as troublesome. Automatic features for adding metadata and information about the context are needed. Various options for this could be used: totally manual user initiated metadata, the system suggests and the user confirms, or totally automatic device input metadata (e.g. as tags). Annotations are important in making the memory more alive. However, with the troublesome text input features in mobile phones, other kind of annotating features are needed.

5. Let the user control what kind of metadata is added and shared. Most of the various kinds of metadata types are seen as interesting and useful: location, people, bio information, textual descriptions, near-by points of interest, communications history etc. The content itself should include metadata enough to tell the story and evoke users’ own memories about the matter at hand. The role of metadata emphasizes when viewing other people’s memories. However, not all of these are wished to be captured in all situations or shared with other people. For controlling what metadata is captured and shared proper information about what metadata is attached, or maybe different privacy levels in capturing and sharing are required

6. Allow the users to define the other users’ accessibility level for their content Different memories are wished to be shared with different people. For example, the most private or intimate memories are shared only with the closest people while, for example, the summer holiday photos may be shared publicly in the Internet. Even in the same event recordings for different recipient groups may be captured. This same ideology applies also to sharing the metadata in the memories. A three-level division for accessibility levels came out several times: private / family and friends / public. Users wanted to keep the options simple instead of having more accessibility levels in, for instance, the circle of friends. Most memories are suitable to be shared with at least the


closest friends and/or family. The accessibility level should also be able to be changed in order to check what others see and what not.

7. Provide a favorable platform for communal services Users value the input from other users a lot. The input can be comments and annotations on digital memories, or interaction around the content when reminiscing about the things captured in the recordings. Users enjoy sharing their memories with others and possible attention through it. Thus, the system should emphasize the communal sharing and interaction features. As many memories about events with a group of close people are regarded as collective memories, there should be features for combining and creating collages from several users’ recordings.

8. Both browsing and searching have to be supported The habits considering browsing and searching polarized strongly in this study. For some users browsing is an inseparable part of memorizing and even when searching for a certain content object, it is easier to find it by browsing the digital memories in chronological order. However, others tend to frequently search for certain memories and recordings, and they do it by searching with a keyword or tag, or by setting a bookmark to it. The users’ viewing behavior depends also on how often and what memories are viewed. More efficient and natural ways to search photos, for example by showing a suchlike photo or by rich context information tags, are needed.

9. Make organizing according to time, event and place very easy The user studies pointed out that when browsing or searching memories it is most natural to view them in chronological order. People remember the timeline in a continuum of various events. For example, my birthday → visiting an old friend → at the grandparent’s → bicycle accident, and so forth. Location and the name of the event serve well especially when searching for certain photo or other recording. Moreover, browsing public content by location would provide totally new points of views into one’s recordings. Also several other interesting ways to put the memories in order came out in the study, so actually there could be an option to browse the memories by any information type.

10. Let the visual data play the leading role A human being is by nature a visual-oriented being, and this came up in the user study quite strongly, too. Thus, the visual information in memories should be highlighted. Some users preferred the photo as large as possible and without any other information attached. They pointed out that the photo itself is the best to evoke own memories and feelings. In addition, there should be various viewing levels for viewing the content with and without the metadata and annotations, and see varying amount of content objects at one time.


11. Provide some kind of features to go through the recent events. The idea of being able to browse through the most recent memories came up with several users. Despite there are ways to do this in current systems too, the users emphasized that it has to be very quick and easy. Especially in the capturing device’s UI, it would be most useful to easily and quickly see how the latest recordings came out, and discard part of them if necessary. Often, several captures are taken from the same object or situation, and then choosing for the best right after capturing is needed. This is especially important when there is little storage space or a poor quality capturing device. Overall, most of the drivers can be used in design of other content management systems and even also other mobile systems. Of course, with such small group of user representatives the results can not be generalized to represent larger population. Thus, the drivers should be regarded as more of guidelines than rules for UI design. Moreover, further interpretation is wise to perform when adapting these. However, when designing the prototype in this case, the drivers could be used as a checklist and an idea source.

7.5. Resulted Prototype

The resulted prototype follows the rules and guidelines that have come out in the design drivers, user needs and consolidated models. It includes some of the design ideas emerged during the user study. The prototype is partly defined only on the concept level, partly very specifically. Because of limited resources, the design phase was limited to design only the mobile UI and exclude the PC and other possible data terminal UIs. Nevertheless, the system is designed rather extensively on concept level, and the features and ideology described next will affect the functionality of PC and other possible UIs, too. This section will introduce only the most novel aspects in the prototype, and the most traditional features, such as capturing, are left out. The screens shown here are screen mock-ups to tell the idea of each screen and to visualize certain features. In addition to the description in this thesis, a user interface specification document from the most central screens and menus in the mobile UI was created. The resulted prototype is an application for capturing memories, sharing them, and automatically organizing them. The ideology is that the memories can be any content put in the system, for example photos, videos, audio, communication history, location & biometric information, blogs or other writings. The content is mostly user-created but also commercial content (e.g. music, games and postcards) can be part of it. Although often valuable as memories and interesting to be shared, shared commercial content becomes problematic considering copyright issues. Visualization paradigms and ways to manage various kinds of content is here limited only to the traditional self-created content, such as photos, videos, annotations and other visual media. However, the ideology in the prototype involves also these because their influence on evoking memories is significant.


The prototype is aimed at the same kind of technology innovators and early adopters that were inquired in the contextual inquiries. This group of users who are keen on mobile photographing and novel technological innovations was a natural choice as it was studied during the entire UCD process. In addition, the results of the user study were not seen to represent the general public. User interface-wise the prototype is designed for current series 60 phones considering the screens size, input devices and contemporary UI design paradigms. However, certain technical aspects are not taken into consideration, such as connectivity (data transfer rates & protocols), capturing and sensor technology and processing power. Nevertheless, while having both novel UI ideas and extensive user study results, a suitable balance between user pull and technology push could be found in the design. The system is divided in mobile and PC interfaces but it can be used with either. Hence, the use context can vary a lot: content is captured, shared and viewed in diverse situations and environments, the length of use varies, and different devices are used for similar actions. Thus, a wide set of features was designed to be included in both terminals as there should be option to perform various actions whenever and wherever. The content is stored in the Internet and it’s available for both data terminal equipments without synchronizing. Also other devices than PC or mobile phone could be used to capture and browse the memories, for example a PDA, TV or various sensor devices. The designed prototype could be thought as a replacement of current mobile phones’ camera and gallery applications. The mobile phone is used mostly for capturing and sharing the memories while the PC application is used for browsing, organizing, editing and annotating. Figure 7.5. describes the prototype’s main functionalities in both platforms.

Figure 7.5. The prototype outline. The prototype focused on the Mobile Phone user interface.


Capturing the media includes also features for adding context data to the memories. Providing other than mere visual or audio data makes the memory of the object more alive. The context data can be captured continuously during the event (described in Section 7.5.4) or by attaching available context data to the single recordings at the moment of capture. The context data can be, for example, location, bio-signals, device information, weather, or near-by resources and people. In addition, users may add their own tags to the memory items and annotate them in order to enrich the shared memory. This division does not introduce much novelties compared to current systems as the design is strongly based on user needs and current behavior. Introducing too many novel features would not be beneficial considering the success of the system as an end-user product. The most novel feature is the event-based ideology which is presented next.

7.5.1. Event-Based Approach The results of extensive UCD process lead us into creating an event-based functioning ideology. The initial idea of event-based management is based on the study findings that the capturing and use of digital memories is strongly based on real-life events. One burst of recordings can easily be identified as one event, and it is natural to organize the recordings in entities according to real life events. Event-based functioning means that the users can create events in their mobile phones to store all the recordings from the corresponding real-life event in one place: the Internet. Users can easily share recorded memories by sharing the event to a certain user or a group of users. Moreover, the event-based ideology is natural in organizing and viewing the content. This idea of event-based content management is a fundamental thought in the design. Using it affects rather much of the functionality of the rest of the prototype.

Create new event

OK Back

3G

Continuous metadata >

Use preset settings >Name *Event starts >Event stops >Invite users >Sharing items during the event >

Add event tags

Create new event

OK Back

3G

Continuous metadata >

Use preset settings >Name *DescriptionEvent starts >Event stops >

Sharing items during the event >

Add event tags

Invite users >Existing groups >Choose from phone book Find near-by usersAdd new userEdit invited >

□ Family□ Hockey team□ Prague 06

Add event description

Figure 7.6. Left: the event creation screen, Right: the submenus under ‘Invite users’. The event is thought to contain all the material relevant to the memory of the corresponding real-life event. As the modern mobile phones are capable of capturing diverse metadata and PC platform is excellent for managing versatile data types,


creating events to integrate all the data becomes rather straightforward. The event can be thought as a certain period of the user’s time that has common characteristics, for example a party, a vacation or a concert. The users’ actions, needs for capturing and interaction with content remain mostly constant during the event. Also the people related to the event and interacting with the user remain mostly same. Hence, various settings can be given to the event to manage the behavior of the event. The event can utilize settings from previous and template events in which case the event is very quick to create; only the name of the new event is required. Users can create the event before the actual capturing event takes place, during it or even afterwards. The recordings captured during the duration of the event are automatically organized under the event. Several options to define the duration of the event are provided in order to support flexible management of the events. The event can be set to start/stop on certain moment in time. It can be manually started/stopped in the event menu. It can start at the moment when something is captured the next time after creating it, and stop when a given time has passed without any capturing. The concept of event varies between users. Others regard only large immemorial events to be events while others see even the small almost mundane one-day happenings as events (e.g. a night with friends). The event ideology in this prototype does not bound the lowest limit of an event other way than requiring actions to create it. The Collective Aspect in the Events By utilizing Internet as media for easy sharing, collective use of the media content is emphasized in this prototype. Creating events makes it easier to share captured media to different groups. A group of users can be formed to be invited to the event. Every invited user is allowed to view others’ recordings and share their own. Also, different users’ memories from the same real-life event can be merged as collective events. This helps collecting all the event data into one collective memory, and the collective role of memories becomes emphasized. The group members’ personal recordings from the same event might differ a lot from each other. Thus, the collective memory becomes a dynamic collage of different private memories. Even small pieces of personal recordings shared with others might change others’ mental impression of the event. Users manifested also that it is interesting to keep both the private and the collective version of the event. This way it is possible to reminisce about the event from only personal point of view or by using all the collective data. By making the memories collective the ownership of the memories gets vacillated but, on the other hand, allows multi-person interaction with memories starting from the very creation of them. Involving the collective aspect of memories already right after capturing enables the harnessing of collective intelligence and motivates the users to annotate the memories in early phase.


During the event, sharing is either automatic by sharing all the recorded material or manual, in which case the user chooses the sharable recordings any time after capturing. When manual, sharing is done via the menus one by one or in groups of content. Sharing either way helps to interact with both the co-located and remote users.

Figure 7.7. The various user roles when using the collective events. The event-based ideology was designed for close groups that have mutual trust. The starting point is that every invited user is equal. Thus, only 2 different access right levels are needed in the mobile UI (the event’s creator and the invited users). Figure 7.7 illustrates the simple roles in the event. The owner – the user who created the event – has slightly wider rights by being the only one who can delete other users’ shared content, and thus act as some sort of group administrator. Viewer roles can be granted in the PC user interface by, for example, using existing tools for public sharing or maybe with a similar system, too.

7.5.2. Browsing the Memories Event-based automatic organizing makes it easier to browse media in natural entities. Users tend to view the memories mostly in chronological order, and moreover, one real-life event at a time. Being able to view all the recordings related to an event makes the user’s own memory more alive. In addition, it is vital to support browsing the media from different viewing levels, because browsing can be done either for close viewing the recordings or quick searching for a certain recording. The prototype mixes ideas from browsing and searching in one feature. Browsing the content can be filtered by viewing 1) only user’s own recordings, 2) only others’ content, 3) only the yet unseen content, or 4) all content. This provides options to quickly view different users’ shared content, the newest content or to search for certain content by browsing a limited set of content. Thus, the users may view both own or other’s points of view from the same event. Also other filtering rules could be created, for example based on tags or certain metadata types.


The content can be sorted in various orders in case the chronological event-based order does not suit the usage situation. The content can be viewed in order by 1) time, 2) event name 3) location (e.g. in alphabetical order) 4) content size. Of course, also other orders can be useful, and they can be based on any common metadata type that can be put in logical order, such as user ratings as stars. According to contextual inquiries, organizing content based on user set tags, time or place is most natural for users.

Figure 7.8. Two levels of view. Left: Events listed. Right: content in one event listed as thumbnails. The browsing filter and current sorting told in the top bar.

Figure 7.9. Two more levels of view. Left: One content object (here a photo) with part of its metadata. Right: content’s details and all metadata attached. Various viewing levels were designed to support viewing of memories in large entities to form a general picture of the event, and secondly viewing of the memory items close and in high detail level,. First, the user has to see the events in a list to view one’s


timeline at a general level (Figure 7.8, left). Secondly, the user wants to see several items at a time at the event level (Figure 7.8, right). Moreover, the single item is wished to be viewed very detailed and, when talking of photos, as large as possible (Figure 7.9, left). There is also an additional screen is for viewing the content in full screen. Moreover, the visual content may be zoomed in/out for seeing the smallest details in mobile context, too. Finally, the various metadata can be seen in one screen to view the richness of the recording (Figure 7.9, right). In order to let the user freely browse in various viewing levels, the change of level has to be swift, and the technology has to be powerful enough to support this. In this prototype, the NaviUp and NaviDown keys in the Navi™ key are harnessed for changing between the levels. In the contextual inquiries users emphasized the need for quickly viewing the most recent captures. This was taken into account by offering a quick route to the gallery of the most recent captures. In the capture mode, user may press the NaviLeft key to jump in the list of items (Figure 7.8, right) where the browsing filter is then set to ‘All’ and the sorting rule to ‘Time’. Providing this option, the user has a quick route to view the most recent captures in order to, for example, share or delete some of them. Again, sets high standards for mobile technology.

7.5.3. Annotating, Tagging and Rating User-initiated annotations came strongly out during the user studies. This inspired creating various features for annotating the content. The users may annotate each others’ shared content by giving tags, drawing annotation boxes, rating with stars, marking as favorite or adding text comments. The features are broadly shown in Figures 7.10. and 7.11.

Figure 7.10. Drawing an annotation box and writing the comment in it.


The feature for drawing annotation boxes in visual content was inspired by the suchlike feature in Flickr. Adding this feature to mobile UI provides tools for instant annotation of the content. The annotations may be viewed in the content item’s details screens.

Figure 7.11. Left: Rating an item with stars. Right: adding tags to a content item . In addition to capturing context data automatically as metadata, the user may add own tags to single content items or to all the content in the event by giving tags to the event itself. Different tags can be given for content captured by oneself, by others and for other kind of content that is not actually captured, such as links to other documents. Tagging mode (Figure 7.11, Right) is always easily achieved by pressing the ‘Pen’ key in the mobile phone keyboard.

7.5.4. Continuous Capturing Continuous capturing was one of the features explored in our concept. The aim was to design a system that can minimize the lost important moments and that allows users to focus on experiencing the moment rather than dealing with a capturing device. This drove us into designing features that would support user-controlled continuous recording. Thus, the User-Initiated Buffering –solution (see below) was designed. Moreover, a feature for continuous context data capturing (see below) was designed to support at least some kind of truly continuous capturing.. The media captured continuously is video, audio or mere context data. Truly continuous video recording is limited by the fact that the camera is not wearable or convenient to be aimed at the target all the time. This shortcoming could be addressed by using a wearable camera. However, the prototype design was limited to mobile phones as currently, and such external capturing solutions were not included.


User-Initiated Buffering Here, buffering is a capture mode where the user can set the device to record video or audio into a buffer of selected length. When something interesting occurs the user may save the media in the buffer – or a part of it – and continue recording in video mode after that. Optionally, the buffering can be stopped when the contents of the buffer is saved (“Save & continue”). These two options complement each other as sometimes the recording is wished to continue while sometimes the saved clip is wished to be viewed right after saving. Figure 7.12 demonstrates the buffering user interface.

Figure 7.12. Left: the user initiated buffering in stopped mode. Right: buffering running and the quick menu options showing The length of the buffer defines how much of the past can be saved permanently. If the buffered material is not saved, it will be overwritten as the buffering goes along. This solution is most suitable for predetermined situations where only part of the recording is important, but unforeseen (e.g. a sports event). Requiring user’s effort diminishes the idea of truly continuous capturing behind this. However, this solution is a compromise between the concept ideology and users’ real needs and suspicions. Continuous Context Data One way to record data continuously is to record mere context data without any audio or visual media. Leaving the visual information out makes it possible to capture information with less ethical suspicions of capturing people’s privacy data. The continuous context data can be, for example, location, weather, near-by points of interest or personal bio-information. In the prototype, recording context data during the event can be either truly continuous throughout the whole event, or the context data can be attached to a single memory item at the moment of capture. This continuous and semi-continuous metadata can be presented in graphs over time or other visualization methods, and may serve as search keys. The included context data types were chosen


after paper prototype tests. To keep the sharing of recordings simple the objective was to choose such context data types that users are willing not just to capture but to share, too. From the sharing perspective, continuous capturing is promising since the missed moments are usually the basis of the precious memories that would have been shared. The continuously captured context data enriches the life memories for collective use. Users can then further annotate the content item in order to make it as vivid and personal as desired.

7.6. Acceptability of the Prototype

As the UCD process was based on frequent interaction with users, the acceptability of both very tentative ideas and detailed prototype design could be measured throughout the process. However, with a small take from the users, only little quantitative results could be delineated. It should be taken into account that most results here are based on only about 15 users’ opinions. In the second prototype test round, collecting acceptability results of the system was based on using the UI which was presented with mock-up screens. On the final user test round the prototype acted as an ideological tool for capturing and sharing content and the UI was only shortly introduced. This made it possible to gather acceptability notes from both the UI and the system’s ideology in general. However, careful interpretation was needed to make any final conclusions by blending the results of these two different test rounds. The prototype received generally good acceptance feedback from most test users. The user interface itself felt usable and intuitive, most probably because of extensive iteration. Users liked the idea of creating the event and setting parameters for it already before the actual event. Preset settings in event creation and group inviting features were highly appreciated. Users thought that the event settings should be changeable anytime because the time and invited group of the event might change even during it. Furthermore, creating the events afterwards has to be supported in order to be able to organize content in events also in case the event was forgotten to create. The event was also regarded as a, for example yearly or weekly, repeated event that could be stored at certain time intervals (e.g. capturing a photo of oneself every day during one year, or a school photo of every year). For such usage the event should be able to put on pause and retrieved later on. Of course, in this kind of seldom use cases the organizing could also be manual in the PC user interface. Certain features, such as reorganizing events, managing the overlapping and tagging the events’ recordings were regarded mostly as PC UI’s features – just as they are. Only tagging is possible in the mobile UI. Generally, using the prototype matched their expectations based on the first verbal description about the prototype and its ideology.


The sharing instantly during the event was regarded fast and efficient way to share with the selected users. With most users, instant sharing suited their current habits and needs well. About half of users shared manually and half would use the automatic sharing of all recordings during the event. It was also revealed that only few users are willing to share the content in public. Maybe this tells that for most users the initial motivation in sharing in Flickr, or other similar, is actually the sharing to close friends or relatives – not to share the content with everyone. Event-Based Ideology Under Inspection Generally, users thought that the event-based capturing and sharing would raise their use of camera phone and sharing the recordings. They stated that many awkward phases are improved in this prototype, for example, manual sharing and organizing the recordings. Creating the event was regarded useful enough to bother doing it. Most users agreed that the event is worth creating even in small events. For most users the smallest event would be something like a meeting with friends. Users thought that with this system the threshold for capturing gets lower. Thus, people could learn capturing also more mundane and surprising – but still interesting after years – moments, such as learning to swim or seeing an old friend surprisingly, instead of only capturing the pre-planned “traditional” events, such as birthdays, festivals and vacations. Users stated that in unforeseeable occasions it is important to provide an option to quickly create an event by, for example, marking the first recording in the new event. The balance between experiencing the event and capturing and annotating during the event troubled many users. No clear general balance could be found as the attitude is highly dependent on person. Acceptability of Continuous Capturing The idea of capturing continuous metadata during the event received a warm welcome. However, the comments on different metadata types were conflicting. Not surprisingly, location and near-by resources (devices, services) interested most. Capturing user’s bio-signals interested a few of the users but few needs for continuous capturing or personal use cases were identified. Weather data, user or device activity history and news feeds received only little interest. Users manifested the need for testing the data types in a functional prototype before being able to express the final opinions. They emphasized the role of semantic meanings when representing any of the captured metadata. Continuous capturing through buffering – as presented here – polarized the opinions of the users. By some users continuous capturing was seen useful for only limited situations. Most users doubted the need for truly continuous capturing, and liked the idea of semi-continuous capturing by user-initiated buffering. On the other hand, some users expressed the desire for truly continuous capturing. They said that the very idea of automatically capturing moments when something interesting occurs can not require


effort from the users. Some of the test users presented innovative ideas, for example, they brought in ideas inspired by the buffering features in modern digital TV set top boxes and bookmarking of streamed data. Bookmarking the continuous data was liked as an idea since it enables easy finding of the interesting part in a long recording. Users thought the model of user-initiated buffering to be the lightest and most ethical, but at the same time inefficient for saving the usually missed moments. For example, the saving options in this solution require further designing. The system would greatly benefit from having hard buttons on the device that would allow the user to easily save, start and stop any recording. Overall, the idea of capturing the usually missed moments is appealing, but the design requires further effort. Users stated that the control must remain in user’s hands in both capturing and sharing. General Suspicions The biggest suspicions were related to the cost of using the service and how the current technology would serve the implementation. The battery, memory and data transfer consumption of continuous capturing raised contradictory feelings. Generally speaking, the current technology was not regarded good enough for the prototype: mobile phone cameras, screens and input methods are usually insufficient. Because of that several users prefer as little tasks done with the mobile phone as possible, and prefer using a PC and digital cameras. Since our study group included also very technology-oriented users, the current technology’s restrictions in prototype’s feasibility slightly diluted the reception. Knowing the current technology limited some users’ future-geared thinking. To recapitulate, the system can succeed, if it is simple enough, it is easy to transfer recordings, technical restrictions are overcome (photo/video quality, speed of transmission, mobile UI), and the costs of usage are reasonable. Creating the event must be kept easy to bother doing it – and so does the inviting and receiving invitations. Other capturing and viewing devices have to be easily integrated to the system when necessary.

7.7. Design Ideas for the Future

This section describes some of the future redesign ideas regarding the prototype. Most of them were left outside the prototype, once again, because of the limited time resources. Various Solutions for Continuous Capturing Buffer capturing was considered rather inefficient for capturing the moments that – because of quickness or unavailability of capturing devices – usually are missed. Hence, during and after the paper prototype tests diverse ideas for continuous capturing were brought out. The ideas divided into three groups: truly continuous capturing with bookmarking, continuous buffering, and buffering of other than visual information.


When asked, some liberal-minded users got interested in the idea of truly continuous capturing. Although of being rather unethical with the privacy violations, the benefit of capturing all the interesting matters was regarded more important. Such persistent capturing could adapt to current conditions to minimize the amount of data stored. For example, when dark, only audio would be recorded, or when stationary, the frame rate would diminish. Furthermore, user could mark the interesting spots in time with a hard button in the phone. Also certain triggers in the context, such as bio-signals or loud noise, could cause marking of a bookmark. With these bookmarks the interesting events would be easily found. The bookmarks could be annotated, and the bookmarks could be of various levels of importance. An intermediate solution between the truly continuous capturing and buffer capturing would be a continuous buffering where the device would continuously buffer a short amount (e.g. 1 minute) of video. The buffer would be stored by pressing a hard button on the phone. In addition to this, user could also use the user-initiated buffering, for example in preplanned events. Also other types of data that video could be buffered. Buffering bio-information or phone calls could be useful in certain situations. However, not any of these ideas were designed any further although especially the truly continuous capturing was discussed with users. Sub-Events for Further Organizing Event based ideology in the prototype have brought out an idea of sub events. For example, a weeklong vacation could include several sub events that would be located under the vacation event but still form events of their own with individual settings. The sub events’ size would be more suitable for one-time sharing, because users do not want to view too much at one time. The feature of sub events could be designed to be based, for example, on user-set continuous tags. When adding tags to content, there could be a prepared set of tags in order to ease the finding later on. So, instead of free tagging it would be more like adding keywords. For example in Finnish, the words can be in numerous forms, and finding content by certain tag word does not necessarily provide all the relevant results. An Intelligent User Interface To ease the managing of the content, the UI could adapt itself to the amount of content or based on the user’s actions. For example, the thumbnails could scale in size based on how many there are, or the UI could hide unused features and bring forth the most used features. Of course, in such automatic and even surprising activity, the UI actions should be confirmed. While browsing, the user could be suggested where to go next by knowing his and other users’ navigation history, similar content and the yet unseen


objects. As the editing and other post processing is tedious to do for every content object, the UI could learn the users’ preferences and make the editing automatically. The editing would mostly relate to photo processing, and video or audio clipping. The UI could ask the user (e.g. while using the PC UI) to fill in missing information, for example, on the contents of the media file or the names of people around. Moreover, the UI could learn certain semantic meanings related to location and other devices. Thus, the location information would not be only numeric but a name: “home” or “at Smith’s”, and the device information could the name of the device or the owner. Also temporal information would be more natural to represent as semantic meanings: “Wednesday”, “at night” or “the Christmas holidays”. In order to ensure the conservation of certain data, some data could be marked as vital that can not be deleted without a password or some other check.

7.8. Summary of the Results

An extensive set of results was obtained during the research. First, the benchmark revealed the current state of the PCM applications and the problems and flaws that existed. Next, user needs, requirements and attitudes towards content management was investigated in the user needs study. The main results were the important role and various levels of sharing, and the suspicions against continuous capturing. User studies in Contextual Design revealed users’ activities with PCM applications and the sequences and interaction models in them. Among other things, capturing in bursts and the varying needs regarding browsing and sharing could be found. The interpreted results of the user studies became concentrated as design drivers. Partially general guidelines for PCM applications and partially case-specific drivers covered the whole extent of the research so far. The most concrete and extensive results was the designed prototype. The prototype involved ideas of event-based capturing, managing and sharing, user-initiated buffering for continuous capturing, new ideas related to browsing and annotating the memories etc. Although not being implemented, the acceptability of the prototype was regarded rather high. From the set of features that were profoundly evaluated, the event-based ideology received most thanks while the user-initiated buffering was regarded to require further design in order to be sufficient.

95

8. CONCLUSIONS

In this final chapter a brief summary of the thesis and its contents is given, the relevance and novelty value of the study is discussed and certain ideas for future research topics are presented. The key output of this thesis was the designed prototype for mobile memory management and its novelties. They are most useful for the customer company in product development and implementing and commercializing concepts. As another output, the research results of user needs study, Contextual Design models and design drivers can be of use in further research in the science community. Their importance can be seen as essential basic research in the fields of personal content management, persistent capturing and context awareness. They serve as sources for ideas of future research areas and contemporary results of extensive user studies.

8.1. Summary of the Thesis

This thesis has reported the UCD process and its resulted prototype based on the SharMe concept. The UCD process was extensive: almost hundred users involved in it during the whole process. The user study produced various kinds of results. First, understanding about the current SharMe related systems and the underlying problems in them were gained by means of benchmarking. User needs regarding capturing, managing and sharing life memories were gathered by using user field study methods, for instance thematic interviews and focus groups. Then, the users’ activities with two state-of-the-art systems – Flickr and camera phones – were investigated in the Contextual Design phase in both mobile and PC environments. The contextual inquiries aimed at finding out the sequences and interaction with the application and the content, and discovering the problematic phases in them. Chapter 6 elaborated the UCD process in depth. Based on the consolidated user data, design drivers for a new mobile system for personal content management were composed. Then, based on the drivers and design ideas brought out during the user studies, a prototype of the SharMe concept was designed. The prototype incorporated novel ideas from context awareness, continuous capturing via buffer capturing and an event-based approach for using and managing the self-created content. The event-based approach was a fundamental idea in the design as

8. Conclusions 96 it had influence on most functionality in the prototype. The results of the UCD process are more specifically described in Chapter 7. Finally, the prototype’s acceptability was evaluated with users from both the viewpoints of user interface design and suitability for its purpose. Fortunately, the evaluation confirmed some of the research team’s own preconceptions about the acceptability of the prototype. However, it also revealed totally new aspects and needs in the area of memory sharing and management, such as the importance of face-to-face sharing and general suspicions regarding automatic and continuous capturing. At the same time, these were considered as the most surprising results of the user studies. People seem to be rather traditional when it comes to managing precious memories: ethics and conventional habits are followed. The research discovered interesting aspects about capturing and sharing life memories. Sharing memories is most natural for people, and interesting especially when performed face-to-face and telling about and discussing the memories together. People tend to capture memories mostly in real-life events. As the events are usually experienced with a group of close people, it is natural to base the sharing of memories around events and the related group. The various elements in the context could enrich the captured memory but users were not as intrigued about the ideas as the researchers were. Test users agreed on the basic ideology of the developed SharMe prototype. According to users, continuous capturing required most redesigning in the prototype. Event-based ideology can be said to be the most intriguing finding in the research and one of the greatest novelties in the prototype. As the idea came up during the Contextual Design process, it can be said that it would not have been discovered or designed in the prototype without a user-centered approach. In addition, user-initiated buffering was another interesting and novel idea. However, in addition to user studies, the SharMe concept contributed to the emerging of the idea. Without the initial idea of continuous capturing in the SharMe concept, this idea probably would not have been born.

8.2. Discussion

As creating the prototype with high user acceptance was the main object of the study, it can be stated that the goal was reached. The users’ sequence and interaction patterns and design drivers could be drawn but they still acted mostly as an aid in designing the prototype. The universal scientific goal of producing new knowledge was achieved since the user study results both produced new information on Flickr and camera phone usage, and confirmed existing knowledge of, for instance, the users’ needs. By designing a prototype of a new product the new knowledge was immediately adapted. Having no exactly comparable research, the results can not be analyzed against other results of similar research.

8. Conclusions 97 Reliability & Validity In such user-centered research the reliability and validity of results has to be discussed. Other researchers having precise knowledge of the exact questions asked in the various user sessions and with same user groups, the research process could be repeated with producing similar results. Similar conclusions of the results of the user studies could easily be made. When studying such a new phenomenon it is challenging to find the real causalities between causes and users’ actions. The real causes for users’ certain actions or attitudes are hard to discover. Moreover, with such narrow and rather biased (photo-oriented) sample from the population it is impossible to generalize the results to cover whole population. Nevertheless, one argument in favor of generalizing the results to cover larger population would be the extensive acceptability of the implemented prototype. Yet, the prototype only represents one interpretation of the user study results – not the results themselves. The researcher’s prior knowledge easily biases the results. Having a multidisciplinary group of researchers in the analysis phases diminished the influence of single researcher on both gathering the user data and analyzing the results. Thus, the results can be said to be little biased with researchers’ prejudices, own hypotheses or frames of reference. The various slightly overlapping research methods to study, for instance, users’ needs increased redundancy. In addition, using several users in each study phase diminished each user’s influence on the results. Finally, the integrated affinity wall as a representative of all the study phases’ results, served as basis when composing the design drivers. By this generalization the results became more reliable and useful for other design cases, too. To sum up, proper triangulation in study methods and research team raised the reliability of the results notably. By our insight, the Flickr user representatives did not represent the Flickr users as well as they could have. They were much more photo-oriented and technology-oriented than thought in advance. This can be due to several things. First, when recruiting, these users being the most active Flickr users saw the recruiting messages first and they could answer immediately. Moreover, as they were eager to show off their photos rather than sharing only to small group of people, participating in a user study would be the very thing they wanted. Finally, they were interested in developing Flickr as they used it a lot and had already lots of ideas how to perfect it. If the group really did not represent the Flickr users well enough, the interaction and sequence models’ validity may have been reduced. Study Methodology The overall focus of the study was broad but that was necessary when designing a new prototype of such new and wide area of activities. If the functionality of the prototype was decided narrower (e.g. only sharing the memories) or not designed a prototype at all, the focus would have been narrowed down. In the contextual inquiries, the focus

8. Conclusions 98 could have been narrower by focusing only on the sequences and interaction in order not to study the users’ needs again. However, with additional results to user needs it turned out to be useful to study the users’ needs in CIs, too. Actually, we could have followed the users even more because our initial vision on Flickr usage was somewhat different than the actual usage, and the question outline did not cover all areas as well as it could have. The research methods were mostly suitable for this kind of design case, but small improvements could have done in a few phases. First, a more proper literature review in the beginning of the project could have given insight into research methods conducted in other similar studies. Introducing a self-reporting method with longitudinal studies (e.g. diaries) could have brought out new aspects in the usage patterns. Also the fact that user needs studies had not ended before contextual inquiries started created challenges to designing the inquiries as the final results had not been interpreted. Fortunately, this did not become a disadvantageous factor. Contextual Design method proved to be the right choice for the study. Although being a heavy process, it was rather easy to adapt it to our needs. The emphasis on data gathering and analysis methods suited our goals. By integrating interviewing, observation and real-item modeling, contextual design was most suitable for investigating the usage sequences and interaction, and inquiring the needs of the users. However, it was challenging to apply CD when designing applications with no existing comparable system. In the CIs the focus had to be chosen from existing sub-task of suchlike activities. Even though this produced satisfactory results it could have been useful to combine CD with other methods, such as diary like in a study by Carter & Mankoff (2005) or content analysis methods. Having more resources, several things could have done to make the results more profound. Writing user personas (Cooper 1997) in CD to concretize the users’ needs, motivation, common tasks and goals could have been useful in proportioning the research results and eased redesigning the work. Similarly, proper graphic visions and storyboards in redesigning phase could have brightened the ideas in our minds, and discover even more efficient ways to perform the tasks. In addition, they could have been used in prototyping sessions to describe the system to the user. Drawing the UEDs, however, would not have given too much benefit when compared to the time it would have taken. Pair or group testing in the early phase of prototyping could have brought out the interaction aspect better, and the iteration could have been faster with more user notices in a short time. Analyzing the users’ content objects could have brought out various aspects in the content’s point of view, which would have deepened the analysis phase. However, to be able to make proper conclusions of it, quantitative – and time requiring – approach would have been needed. Finally, a more proper heuristic evaluation for the prototype in various phases of the design would have reduced the

8. Conclusions 99 evident usability faults. Thus, the users in the paper prototype test could have concentrated in other things than pointing out these obvious errors. Now, the design progressed according to our own image of a usable user interface and only shortly used Nielsen’s heuristics as reference for design solutions. The Results Revisited When designing the prototype, the needs and current usage patterns of the users were the most focused inputs. As focusing merely to design an ideal application regarding user friendliness and users’ future needs, the commercial and technical aspects were left aside. To make the prototype technically and economically feasible for current technology, certain functionalities (e.g. bio-information, proximity information and web-based storing solution) have to be forgotten. For a commercial product, the gamut of features is currently too wide. The provided features have to be proportioned to the possibilities to gain business incomes with the product. The gathered user data was very extensive but relatively too little time was spent on redesigning the prototype. Hence, the user data should be utilized to create more prototypes and ideas in order to utilize the data enough. In other words, it was the data gathering and analyzing the users’ requirements that was the research team’s expertise – not innovating and creating new commercial applications. The acceptability results here are only based on feedback from a small group of technology innovators and early adopters, and the results should not be generalized too much. While the traditional user attitude, such as face-to-face sharing, still thrives and most users’ habits are not got used to the abundant sharing and browsing of the memories, the eventual acceptability might be very different. To study the definite acceptability, the prototype has to be implemented and a longitudinal study to investigate the usage is required. Often in user-centered design processes great amount of resources are required. However, in scientific research one aim is to make the results more generally applicable. Especially in this case, most results are in such general level that they may be used in other mobile and PC content management application design, too. The reliability is high enough to meet the requirements of design processes. The resulted prototype would have certainly looked very different without the extensive user-centered research. Moreover, the prototype provided several novelty values for designing a mobile content management application. Also the UCD process introduced some new approaches, and involved in user groups that have not been studied like this before. To sum up, the study succeeded well. Studying users’ needs and usage patterns is a relative common and down-to-earth process. Thus, the most challenging aspect was to find the most relevant notes and ideas, handle with the large amount of information to

8. Conclusions 100 analyze it reliably, and to design an advantageous prototype taking all the emerged aspects into consideration.

8.3. Future Work

To take the full advantage of the needs and requirements of the users, extensive progress in the current technology must take place. Certain open issues were left open with the prototype: how to keep the data transfer amounts within moderate values? Does the transfer rate and photo processing speed serve, for example, quick browsing? Some current weaknesses of the mobile phone, such as processing power and transfer rates, will certainly develop rapidly, while the development of some technologies is not that sure. For instance screen and button size, input methods and camera optics will hinder the development of PCM applications in mobile phones. Partly because of the current technical limitations, also in the design of the prototype certain features were forced to be left outside. The Future of Context Awareness As the total concept of SharMe is rather extensive, further research in several fields is required. This study focused on capturing and sharing the memories. One very interesting – yet under-utilized feature in the prototype – is context awareness. Until now, the research around context awareness has been very exploratory and demonstrative. Here, the context awareness is still mostly to offer extra information for the user about the current and past context instead of enhancing the grasp of the device of the use context. Truly context aware systems with triggered functions or adaptation in UI based on the current context offer a totally new world for, for example, continuous capturing and location aware features. For example, various capturing, browsing or sharing rules in various contexts and intelligent system to know which content other people around you have not yet seen. Such automating functions have to be designed with carefully taking the users into consideration. Considering continuous capturing, it will be essential to know what things in the context can be used as triggers to evoke recording actions. Moreover, what context data should act as triggers and what more like mere metadata, what context data should be recorded or shared in various situations and how precisely, and what kind of visualization paradigms for various kinds of metadata and context data types there could be? Consequently, context data and its usage as metadata or triggers for adaptable user interfaces require further extensive studies to find favor with users. Continuous Capturing Prospects As the continuous capturing makes it possible to capture surprising everyday moments, the capturing does not occur anymore in bursts. Proper sharing features for this mundane data should be designed. For example, automatic blogging, real-time sharing, or long time events between small groups of people where they would automatically

8. Conclusions 101 share their surprising moments in real-time with each other and the material would be gathered in one long “everyday-event”. Furthermore, the specific implementation of truly continuous capturing (running from day to day) has to be well designed, since truly continuous capturing raised suspicion and ethical questions within both the user needs research and contextual inquiries. For this kind of persistent capturing, some kind of wearable solutions are necessary. In addition, further design of the buffer recording is required as it is seen as a second class solution for continuous capturing. The Outlook for Event-Based Content Management The collective nature of memories requires further research with users. Regarding the popularity of such collective prototype, it would be worthwhile to know how different communities regard their collective data and how is it used. Again, a longitudinal study with a functional prototype would be best to reveal the real usage patterns regarding capturing, sharing, annotating and using the collective content. The needs regarding collective usage of the memories might vary from the needs regarding personal use. Also, the temporal aspect in sharing (real-time vs. sharing later) would be useful to study in order to understand what kind of features are needed for real-time sharing, viewing and annotating. Related to this prototype, the temporal behavior of the events requires further designing. How do the pause and stop mode work with events? How to easily change the focus between events? Can there be several events on at a time? Moreover, the idea of sub events interested both us and the users. Again, several questions emerge. How to visualize the nested events? What kind of automation there could be in the creation of sub events (e.g. based on time)? To recapitulate, because of the wide scope of the study, it left several interesting questions open to be further studied. The SharMe concept involved continuous and automatic capturing, context awareness and life memory management, all of which are novel ideas mostly discussed in the scientific forums – not much implemented in products. This project could only open the door to these intriguing fields but not concentrate well enough on any of the matters. In such research areas that concern automation of user tasks, sensitive matters and privacy issues, the UCD approach will be essential in order to obtain results that can be utilized in successful future products.

102

REFERENCES Aaltonen, A. 2007. Facilitating Personal Content Management in Smart Phones. Doctoral thesis. University of Groningen, Netherlands, 2007. 171 p. Anderson, J. 1976. Language, Memory and Thought. Erlbaum Lawrence. 546 p. Baeza-Yates, R. & Ribeiro-Neto, B. 1999. Modern Information Retrieval. ACM Press & Addison-Wesley, New York, NY, USA. 513 p. Bevan, N. & Bogomolni, I. 2000. Incorporating User Quality Requirements in the Software Development Process. 4th International Software & Internet Quality Week Conference (QWE2000). Brussels, Belgium, 20-24 November 2000. Serco Ltd. Beyer, H. & Holtzblatt, K. 1998. Contextual Design. Defining Customer Systems. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA. 472 p. Bush, V. 1945. As We May Think. The Atlantic Monthly, 176(1), July 1945. pp. 101-108. Carmel, E., Whitaker, R., George, J. 1993. PD and Joint Application Design: A Transatlantic Comparison. Communications of the ACM. Vol. 36, Issue 6 ACM Press, New York, NY, USA. pp. 40-48. Carter, S. & Mankoff, J. 2005. When Participants Do the Capturing: The Role of Media in Diary Studies. Proceedings of the SIGCHI conference on Human factors in computing systems. Portland, Oregon, USA. ACM Press, New York, NY, USA, 2005. pp. 899-908. Cockton, G. 2006. Designing worth is worth designing. Proceedings of the 4th Nordic conference on Human-computer interaction: changing roles. Oslo, Norway, 2006. ACM Press, New York, NY, USA. pp. 149-160. Cooper, A. 1997. The Inmates are Running the Asylum: Why High Tech Products Drive Us Crazy and How to Restore the Sanity. Sams Publishing. 288 p. Czerwinski, M. et al. 2006. Digital Memories in an Era of Ubiquitous Computing and Abundant Storage. Communications of the ACM. Vol. 49, Issue 1. January 2006. ACM Press, New York, NY, USA. pp 44-50.

References 103 Dahlbäck, N., Jönsson, A., Ahrenberg, L. 1993. Wizard of Oz Studies: Why and How. Proceedings of the 1st International Conference on Intelligent User Interfaces. Orlando, Florida, United States, 1993. ACM Press, New York, NY, USA. pp. 193-200. Davis, M. et al. 2005. MMM2: Mobile Media Metadata for Media Sharing. 2005. CHI '05 extended abstracts on Human factors in computing systems. Portland, OR, USA, 2005. ACM Press, New York, NY, USA. pp. 1335-1338. Dey, A. & Abowd, G. 1999. Towards a Better Understanding of Context and Context-Awareness. GVU Technical Report GIT-GVU-99-22, College of Computing, Georgia Institute of Technology. Dey, A., Abowd, G., Wood, A. 1999. CyberDesk: A Framework for Providing Self-Integrating Context-Aware Services. Knowledge-Based Systems, Vol. 11, Issue 1. Elsevier Inc. pp. 3-13. Dey, A. 2001. Understanding and Using Context. Personal & Ubiquituos Computing, Vol 5, Issue 1, Feb. 2001, Springer London, UK. pp. 4-7. Dumais, S. et al. 2003. Stuff I've seen: a system for personal information retrieval and re-use. Proceedings of the 26th annual international ACM SIGIR conference on Research and development in information retrieval. Toronto, Canada, 2003. ACM Press, New York, NY, USA. pp. 72-79. Fetterman, D. 2007. Ethnography: Step by Step. Sage Publications, 2nd ed. 176 p. Frochlich, D. et al. 2002. Requirements for Photoware. Proceedings of the 2002 ACM conference on Computer supported cooperative work. New Orleans, Louisiana, USA, 2002. ACM Press, New York, NY, USA. pp. 166-175. Frohlich, D. 2004. Audiophotography: Bringing photos to life with sounds. Springer-Verlag New York, Inc. Secaucus, NJ, USA. 242 p. Gargi, U., Deng, Y., Tretter, D. 2002. Managing and Searching Personal Photo Collections. HP Labs Technical Report HPL-2002-67, 2002. Gemmell, J. et al. 2002. MyLifeBits: Fulfilling the Memex vision. Proceedings of ACM Multimedia’02. Juan-les-Pins, France, 2002. ACM Press, New York, NY, USA. pp. 235-238. Gemmel, J. et al. 2004. Passive Capture and Ensuing Issues for a Personal Lifetime Store. Proceedings of the the 1st ACM workshop on Continuous archival and retrieval

References 104 of personal experiences, CARPE ’04. New York, USA, 2004. ACM Press, New York, NY, USA. pp. 48-55. Graham, A., Garcia-Molina, H., Paepcke, A. and Winograd, T. 2002. Time as essence for photo browsing through personal digital libraries. Proceedings of the 2nd ACM/IEEE-CS joint conference on Digital libraries. Portland, Oregon, USA, 2002. Press, New York, NY, USA. pp. 326-335. Haikala, I. & Märijärvi J. 2000. Ohjelmistotuotanto, 7th edition. Satku – Kauppakaari Oy. 414 p. Häkkilä, J. 2006. Usability with Context-aware Mobile Applications – Case Studies and Design Guidelines. Doctoral thesis. University of Oulu, Finland, 2006. 77 p. Hayes, G. et al. 2004. The Personal Audio Loop: Designing a Ubiquitous Audio-Based Memory Aid. Mobile Human-Computer Interaction, MobileHCI 2004. Glasgow, Scotland, 2004. Springer Berlin, Germany. pp 168-179. Healey, J. & Picard, R. 1998. Startlecam: A Cybernetic Wearable Camera. 2nd International Symposium on Wearable Computers. Pittsburgh, PA, USA, October 1998. IEEE Computer Society, Washington, DC, USA. pp. 42-49. Hodges, S. et al. SenseCam: A Retrospective Memory Aid. Proceedings of the 8th International Conference on Ubicomp 2006. Orange County, CA, USA, September 2006. Springer-Verlag Berlin, Heidelberg, Germany. pp. 177-193. Holtzblatt, K., Wendell, J., Wood, S. Rapid Contextual Design. A How-To Guide to Key Techniques for User-Centered Design. 2005. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA. 318 p. ISO CD 9241-11. Guidelines for specifying and measuring usability. 1993. ISO/IEC 13407. Human-Centred Design Processes for Interactive Systems. 1999. ISO/IEC TR 18529. Human-Centred Lifecycle Process Descriptions. 1999. Jacucci, G., Oulasvirta, A., Salovaara, A., Sarvas, R. 2005. Supporting the Shared Experience of Spectators Through Mobile Group Media. Proceedings of the international ACM SIGGROUP conference on Supporting group work. Sanibel Island, Florida, USA, 2005. ACM Press, New York, NY, USA. pp. 207-216. Järvinen, P., Järvinen, A. 2004. On Research Methods. Opinpaja Oy, Tampere. 204 p.

References 105 Jokela, T., Iivari, N., Matero, J., Karukka, M. 2003. The Standard of User-Centered Design and the Standard Definition of Usability: Analyzing ISO 13407 against ISO 9241-11. Proceedings of the Latin American conference on Human-computer interaction. Rio de Janeiro, Brazil, 2003. ACM Press, New York, NY, USA. pp 53-60. Kaasinen, E. 2005. User acceptance of mobile services – value, ease of use, trust and ease of adoption. Doctoral thesis. VTT Publications 566, Finland, 2005. 151(+64) p. Kiljander, H. 2004. Evolution and Usability of Mobile Phone Interaction Styles. Doctoral thesis. Helsinki University of Technology. 234 p. Kirk, D., Sellen, A., Carsten, R., Wood, K. 2006 Understanding Photowork. Proceedings of the SIGCHI conference on Human Factors in computing systems. Montréal, Québec, Canada, 2006. ACM Press, New York, NY, USA. pp. 761-770. Koolwaaij, J. 2006. Context Watcher - Sharing Context Information in Everyday Life. Proceedings of the IASTED conference on Web Technologies, Applications, and Services. Calgary, Alberta, Canada, 2006. ACTA Press. Korkea-aho, M. 2001. Location Information in the Internet. Licentiate’s thesis. Helsinki University of Technology. 80 p. Koskinen, I., Kurvinen, E., Lehtonen, T. 2002. Mobile Image. Edita Publishing Inc., IT Press, Helsinki, Finland. 160 p. Kujala, S. & Kauppinen, M. 2004. Identifying and Selecting Users for User-Centered Design. Proceedings of Nordic Conference on Computer-Human Interaction. Tampere, Finland, 2004. ACM Press, New York, NY, USA. pp. 297-303. Lehikoinen, J., Aaltonen, A., Huuskonen, P., Salminen, I. 2007. Personal Content Experience. John Wiley and Sons Ltd. 382 p. Lucas, P. 2001. Mobile Devices and Mobile Data – Issues of Identity and Reference. Special issue of Human-Computer Interaction, volume 16, 2001. Lawrence Erlbaum Associates, Inc. pp. 323-336. Marshall, C. & Rossman, G. 1998. Designing Qualitative Research, 3rd Ed. Sage Publications, London, UK. 240 p. Mäkelä, A., Giller, V., Tscheligi, M., Sefelin R. 2000. Joking, Storytelling, Artsharing, Expressing Affection: a Field Trial of How Children and Their Social Network Communicate with Digital Images in Leisure Time. Proceedings of the SIGCHI

References 106 conference on human factors in computing systems CHI’00. The Hague, Netherlands, April, 2000. ACM Press, New York, NY, USA. pp. 548-555. Michahelles F., Samulowitz, M. 2002. Smart CAPs for SmartIts – Context detection for Mobile Users. Personal and Ubiquitous Computing, Vol. 6, Num. 4, Sep. 2002. Springer London, UK. pp. 269-275. Nielsen J. 1994. Heuristic Evaluation. Usability Inspection Methods. John Wiley & Sons Inc., New York, NY, USA, 1994. pp. 25-62. Okabe, D. Ito, M. 2005. Everyday Contexts of Camera Phone Use: Steps Towards Technosocial Ethnographic Frameworks. Forthcoming in Höflich, Joachim & Hartmann, Maren. Mobile Communication in Everyday Life: An Ethnographic View. Frank & Timme, Berlin, Germany. pp. 79-102. Pascoe, J. 1998. Adding Generic Contextual Capabilities to Wearable Computers. 2nd International Symposium on Wearable Computers, Pittsburgh, PA, USA, October 1998. pp. 92-99. Pascoe, J., Ryan, N., Morse, D. 1999. Issues in Developing Context-Aware Computing. Proceedings of the International Symposium on Handheld and Ubiquitous Computing. Karlsruhe, Germany, September 1999. Springer Berlin, Germany. pp. 208-221. Preece, J., Rogers, Y., Sharp, H. 2002. Interaction Design: Beyond Human-Computer Interaction. John Wiley & Sons, New York, NY, USA. 519 p. Redmond-Pyle, D. & Moore, A. 1995. Graphical user interface design and evaluation guide (GUIDE): A practical process. Prentice Hall PTR, London, UK. 300 p. Rogers, E. 2003. Diffusion of Innovations, 5th Ed. The Free Press, New York. 512 p. Rosson, M. & Carroll J. 2001. Usability Engineering: Scenario-Based Development of Human-Computer Interaction. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA, 2001. 422 p. Saffer, D. 2007. Designing for Interaction. Creating Smart Applications and Clever Devices. New Riders, Berkeley, CA, USA. 231 p. Sarvas, R. 2006. Designing User-Centric Metadata for Digital Snapshot Photography. Doctoral Dissertation. Department of Computer Science and Engineering, Helsinki University of Technology, Finland.

References 107 Sarvas, R., Herrarte, E., Wilhelm, A., Davis, M. 2004. Metadata Creation System for Mobile Images. Proceedings of the 2nd international conference on Mobile systems, applications, and services. Boston, MA, USA, 2004. ACM Press, New York, NY, USA. pp 36-48. Sarvas, R., Oulasvirta, A., Jacucci, G. 2005. Building Social Discourse Around Mobile Photos - A System Perspective. Proceedings of the 7th international conference on Human-computer interaction with mobile devices & services. Salzburg, Austria, 2005. ACM Press, New York, NY, USA. pp. 31-38. Schilit, B. & Theimer, M. 1994. Disseminating Active Map Information to Mobile Hosts. IEEE Network, Vol. 8, Issue 5. pp. 22-32. Schilit, B., Adams, N., & Want, R. 1994. Context-Aware Computing Applications. IEEE Proceedings of the 1st International Workshop on Mobile Computing Systems and Applications, Los Alamitos, CA, USA, 1994. pp. 85-90. Sinkkonen, I., Kuoppala, H., Parkkinen, J., Vastamäki, R. 2006. Käytettävyyden Psykologia. Edita Publishing Oy, Finland. 334 p. Smeulders, A. et al. 2000. Content based image retrieval at the end of the early years. IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 22, Issue 12. pp. 1349-1380. Sorvari A. et al. Usability Issues in Utilizing Context Metadata in Content Management of Mobile Devices. Proceedings of the third Nordic conference on Human-computer interaction NordiCHI '04. Tampere, Finland, 2004. ACM Press, New York, NY, USA. pp. 357-363. Strauss, A. & Corbin, J. 1990. Basics of qualitative research: Grounded Theory Procedures and Techniques. Sage Publications, Newbury Park. 272 p. Tamminen, S., Oulasvirta, A., Toiskallio, K., Kankainen, A. 2004. Understanding mobile contexts. Personal and Ubiquitous Computing, 8, Num. 2. Springer London, UK. pp. 135.143. Van House, N. et al. 2005. The Uses of Personal Networked Digital Imaging: An Empirical Study of Cameraphone Photos and Sharing, CHI 2005. Portland, Oregon, USA, April 2005. ACM Press, New York, NY, USA. pp. 1853-1856.

References 108 Wilhelm, A. et al. 2004. Late breaking result papers in Conference on Human Factors in Computing Systems CHI ’04. Vienna, Austria, 2004. ACM Press, New York, NY, USA. pp. 1403-1406. Vredenburg, K., Mao, J., Smith, P., Carey, T. 2002. A Survey of User-Centered Design Practice, Proceedings of the SIGCHI conference on Human factors in computing systems: Changing our world, changing ourselves. Minneapolis, Minnesota, USA, 2002. ACM Press, New York, NY, USA. pp. 471-478. Väänänen-Vainio-Mattila, K. & Ruuska, S. 2000. Designing mobile phones and communicators for consumer’s needs at Nokia. ACM Interactions, Sep. 1999. Morgan Kaufmann Publishers. pp. 169-204.

109

WEB REFERENCES Amazon [WWW]. [Cited 18/03/07]. Available at: www.amazon.com. Aukea. [WWW]. [Cited 22/03/07]. Available at: http://www.aukea.net. Berners-Lee, T. Metadata Architecture. [WWW] [Cited 13/05/07] Available at: http://www.w3.org/DesignIssues/Metadata.html Context Watcher. [WWW]. [Cited 21/03/07]. Available at: http://www.lab.telin.nl/~koolwaaij/showcase/crf/. Del.icio.us [WWW]. [Cited 30/03/07]. Available at: http://del.icio.us. EXIF metadata [WWW]. [Cited 10/02/07]. Available at: http://www.exif.org. Flickr. [WWW]. [Cited 15/03/07]. Available at: http://www.flickr.com. Flipper [WWW]. [Cited 29/03/07]. Available at: http://www.flipper.com. Futurice [WWW]. [Cited 29/03/07]. Available at: http://www.futurice.com. Habbo Hotel [WWW]. [Cited 30/03/07]. Available at: http://www.habbo.com. Hewlett Packard: Photographic memories. [WWW]. [Cited 10/04/07]. Available at: http://www.hpl.hp.com/news/2004/jan-mar/casualcapture.html. Jaiku [WWW]. [Cited 29/03/07]. Available at: http://www.jaiku.com. Jeff Croft [WWW]. [Cited 29/03/07]. Available at: http://www2.jeffcroft.com/blog/2006/sep/20/personal-content-management. Korkea-Aho, M. Context-Aware Applications Survey. 2000. [WWW]. [Cited 10/01/07]. Available at: http://users.tkk.fi/~mkorkeaa/doc/context-aware.html. LifeBlog. [WWW]. [Cited 21/03/07]. Available at: http://r2.nokia.com/nokia/0,,71739,00.html. Meaning [WWW]. [Cited 02/04/07]. Available at: http://meaning.3xi.org. Multiply [WWW]. [Cited 03/04/07]. Available at: http://www.multiply.com.

http://www.w3.org/DesignIssues/Metadata.html

Web References 110 MySpace. [WWW]. [Cited 15/03/07]. Available at: http://www.myspace.com. Orkut. [WWW]. [Cited 15/03/07]. Available at: http://www.orkut.com. Pandora [WWW]. [Cited 18/03/07]. Available at: www.pandora.com. Second Life. [WWW]. [Cited 15/03/07]. Available at: http://secondlife.com. SharMe project [WWW]. [Cited 30/03/07]. Available at: http://research.nokia.com/research/projects/sharme/index.html. Youtube. [WWW]. [Cited 15/03/07]. Available at: http://www.youtube.com.

111

INTERNAL REFERENCES Lehtonen, M. & Soronen, H. 2007. SharMe user needs study summary v1.0. 27.02.2007. Institute of Human-Centered Technology, Tampere University of Technology Pavel, D. 2006. SharMe project plan v1.3. 22.5.2006. Nokia Research Center, Software Technologies Laboratory.

1(3)

APPENDIX A. SUMMARIZED QUESTION OUTLINE FROM FLICKR CONTEXTUAL INQUIRIES Flickr – A Compressed Outline of Inquiry Questions (Translation from Finnish) User description: (age, sex, education/profession, character, family) Warm-up questions: For what do you use Flickr? How are your friends/family related to your Flickr usage?

• Do you use Flickr alone or in company? • With whom do you share photos? • From whom do you receive photos? • How do you communicate with your friends and family?

How are your hobbies related to Flickr usage? Do you know other Flickr users in real life? What motivates you to share (friends, oneself)? For how long / how much have you used Flickr? Do you use other applications for managing, editing or sharing memories? How much have you used the following devices/systems?

• Camera phone • Digital camera • Film (traditional) camera • Video camera • GPS-device • Wrist top computers/heart rate monitors

How much do you capture photos with

• Camera phones • Film cameras • Digital cameras

In what kind of situations? What kind of photos do you transfer to Flickr? What else do you do with your photos (print, keep in a shoebox, organize in albums...)?

Appendix A 2 (3) Observing the user in action: Is this a situation where you could use Flickr (a friend visiting)? Try to remember your last instance of use. Repeat what you did then Try to remember another recent use case. Try to repeat what you did then. Logging in and the first actions: Go to Flickr, log in and perform the normal actions you do after logging in Do you use Flickr from several computers? How about mobile devices? Uploading photos and sharing them in Flickr: Upload a few photos and share them publicly/privately Give tags and descriptions for these photos What kind of photos have you uploaded to Flickr? Do you share single photos or larger entities? Are you aware of what photos are share and with whom? Do you show your photos to others by other means? Editing, tagging, commenting and organizing photos: Edit a photo with the normal tool you use for editing Give tags and descriptions to some earlier uploaded photos Organize your photos to albums (if suitable) and add descriptions to the albums Comment some of your photos What kind of information do you want to receive from the description of the photo? What kind of tags do you give for the photos? Do you change tags, descriptions or captions later on? How? In what kind of entities do you classify your photos and by what (theme, time etc.)? What kind of editing actions do you perform? What kind of photos do you remove from Flickr? What do you think of the ways to view photos in Flickr? Searching, viewing and commenting others’ photos: Show how you usually browse others’ photos In what kind of ways have you browsed and searched photos? What kind of and who’s photos have you viewed? How do you get information about your friends new photos in Flickr? How do you comment others’ photos?

Appendix A 3 (3) What kind of photos get to your list of favorites? What kind of photos do you view frequently? Communication: How do you interact in various user forums or with other users?

• In what kind of communities/forums are actively involved in? • In what kind of other means do you interact with other users?

How do you classify other users (friend, public etc.)? What kind of information do you wish to see from other users? How does your profile look like to others? Final questions: How fixed are your habits in using Flickr (sequences, navigation etc.)? What features of Flickr would you like to use with a mobile phone?

• How about after next 5 years when most technical restrictions are overcome? What kind of other data (metadata) could the photos include to ease searching, browsing and organizing or to just offer additional interesting information? What elements in the photos trigger your own memories the best? In what kind of entities would you like to organize/categorize your photos? What do you think of the navigation in Flickr? What do you think of the look & feel of Flickr? What kind of shortcuts would you like to have to make your usage more efficient? What kind of real-time instructions would you like to have? What kind of new features or services would you like to have? Would you like to personalize your user interface more? How? What are the best three features in Flickr? What kind of an application could make you forget Flickr and change you to using that? How would you perform the application changing operation? What files, settings and profile information would you like to transfer to the new system?

1(4)

APPENDIX B. CONSOLIDATED MODELS FROM CONTEXTUAL INQUIRIES Flickr – Consolidated Interaction Model

EDITING TOOLPicasa, Photoshop, iPhoto

LOADING TOOLLoading pictures to Flickr

PHOTO GALLERYSharing photos to non-Flickr-users

PHOTOSomething

interesting + annotations

DIGITAL CAMERApreplanned

CAMERAPHONEsnapshooting

SPOUSEViews together (same screen)

FLICKR -userPhotographs

EditsOrganizes

SharesViews photos

Collects feedbackDiscusses about photos

Exploits the photosMotivates others to use Flickr

GOOGLESearching photos

FRIENDS IN REAL WORLDPresent at capturing situation

Share photosWatch photos

FLICKRSharing photos

Receiving feedbackWatching pictures

OrganizingDiscussion forumPlace of storage

Source of informationWay of getting social status

Other FLICKR -usersShare photos

Comment picturesDiscuss

IRC, SKYPE, MESSENGER

EMAILTells new pictures

are availableShares drinking

photos

Photos are too smallUser can make only 3albums for free

It is inconvenient to have photos in many places

photo

Tells if new pictures are available

Tells if new pictures are available

Requests a photo

photo

photo

photo

photo

photo

photo

photo

Watching

photo

Comment / message

Appendix B 2 (4) Camera Phone – Consolidated Interaction Model

Appendix B 3 (4) Flickr – Consolidated Sequence Model of Sharing Photos

Opens the photo storage

Edits with FlickrCropping, color balance

SharesShares most as public

Checks if the photos were uploaded

Goal: to share a photo in FlickrTrigger: The user wants to share photo(s) F: sharing, consolidated model

Edits with an applicationSize, contrast, alignment, name

An pointless phase but necessary because the photos can not be shared to Flickr directly

from the editing application

Requires refreshing the www browser

Opens the sharing tool

Chooses suitable photo(s)

Stores into a temporary directory

SetsTags, description, caption

Editing in two different phases

Each photo separately

Lack of confidence for the application

Appendix B 4 (4) Camera Phone – Consolidated Sequence Model of Capturing Photos

thomas olsson creating, managing and ...olsson/diplomityo_thomas_olsson_sharme.pdfthomas olsson...

Documents