semantic search for media portals
DESCRIPTION
A presentation I have given several times illustrating to non-technical people how the Internet can change information access in media portals. It focusses on the different ways of information organisation and architecture that are possible in digital media because of taking away physical constraints.TRANSCRIPT
Semantische Suche in Medienportalen
Dr. Sebastian SchaffertSalzburg Research / Salzburg NewMediaLab
1
Introduction
2
Sebastian Schaffert
• Doktorat in Informatik, Uni München
• Senior Researcher bei Salzburg Research
• Forschungsgebiete Social Software, Web 2.0 und Semantic Web
• Projektkoordinator des EU-Projekts „KiWi - Knowledge in a Wiki“
3
Salzburg Research• Forschungsgesellschaft des Landes Salzburg
• Fokus auf interdiszipliäre IT-Forschung
• Wissens- und Medienmanagement
• Mobilität und ortsbasierte Dienste
• Bildung und Medien
• E-Culture
• Netzwerktechnologien
4
Salzburg NewMediaLab
• Österreichisches Kompetenzzentrum zu Neuen Medien
• „public private partnership“-Modell mit öffentlicher Kofinanzierung
• Forschung in den Bereichen „Multimediatechnologien“, „Social Software“ und „Semantischen Systemen“
5
Information Organisation
6
video by M. Wesch/YouTube
7
classical paper-based information organisation
is limited by physical constraints and thus
follows a single hierarchy
8
Example: Dewey Decimal System
• developed by US librarian Melvil Dewey
• arranging books in a numerically encoded hierarchical order by subject
9
Figure from Politt & Tinker (2003)
10
but what if your world view does not match Dewey‘s 1930s world view?
11
12
13
This also holds for newspapers!
photo by birdfarm/Flickr
14
15
Computers offer to organise information along multiple dimensions, detached from
physical constraints http://universe.daylife.com/
16
Computers offer to organise information along multiple dimensions, detached from
physical constraints http://universe.daylife.com/
16
Computers offer to organise information along multiple dimensions, detached from
physical constraints http://universe.daylife.com/
16
Different Hierarchies
17
Example: Holiday Photos
18
you could organise as ...
2008
Italy Photos 2008
19
or as ...
2008
ItalyPhotos 2008
20
or even as ...
2008
ItalyPhotos 2008
21
or maybe as ?
2008
Italy Photos2008
22
all this makes sense ...
... to someone
23
but: how many
dimensions are there?
photo by Alex Kessler/Flickr
24
5!(exactly)
25
LocationAlphabetTimeCategoryHierarchy Richard Saul Wurman
Information Designer
26
Location ...
http://tagit.salzburgresearch.at
27
Alphabet ...
http://www.linkedin.com
28
Time ...
http://simile.mit.edu/timeline/
29
Category ...
30
Hierarchy ...
31
What does this mean for News Portals?
32
most existing news portals follow the classical, resort oriented navigation like in
paper-based news - physical limitation lifted to virtual space
33
34
35
• resort = category (sort of ...)
• but: not necessarily topic!
36
• sports • economy• politics• culture• Salzburg
Article on soccer EM could be in ...
37
LATCH in Online News
38
News by Location ...
http://atlas.tagesschau.de
39
News by Alphabet ...
40
News by Time ...
41
sorry, no good example (except resort-based) :-(
News by Category ...
42
News by Category ...
but there is:
http://www.iptc.org
43
so why not offer it for navigation?
News by Category ...
44
News by Hierarchy ...
45
Challenges & Opportunities
46
from big ambitions to realisable goal
47
1. user centred design means „intuitiveness“ of interface
Challenges ...
48
but intuitiveness only exists when facing a bear ...
from: user „randy_harris“ at Flickr
49
otherwise, it is rather patterns and idioms we already know ...bread crumps
tag clouds
home link
dropdown selection
tabs
User Interface ...
50
• when visiting an online news paper, people almost expect a classical navigation structure
• new idioms need to be introduced very carefully (e.g. blog style, ...)
• more complex structures need to be hidden (in salzburg.com: only in search, not in navigation)
User Interface ...
51
2. assuming that editors become „knowledge engineers“ that properly maintain complex knowledge models was unrealistic
Managing Topics ...
52
• need to do as much automatic processing as possible (but this is limited)
• possibility to involve users!
Managing Topics ...
53
Tagging
54
Linking
55
Structuring
from: user „liber“ at Flickr
56
3. integration with other kinds of content beyond news
Integration ...
57
60
Future Content Platforms
61
• Semantic Search (completed 2008): http://search.salzburg.com
• KiWi (platform developed by EU Project):
• Content Integration Framework (2009):integration and connection of different kinds of content
• TagIT (2009):geolocation & social tagging of news and places
Project Deliverables ...
62
keyword-based interface, refine search results by
map, category, time, location
search.salzburg.com
63
DEMO!
http://search.salzburg.com
64
• UI: Ruby on Rails, AJAX• Logic: mostly PL/SQL• DB: PostgreSQL• XML feed of news articles• optimized full-text index, time index,
location index, resort• 700.000 articles
Technology (Productive) ...
65
Data Import ...
Articles(XML)
Geolocation(named entities + geo field)
Fulltext Index(PostgreSQL built-in)
Database(PostgreSQL)
66
• EU project funded under 7th Framework Programme
• 7 partners, 3.8 Million Euro
• develops a platform for „Semantic Social Software“
• builds on the „Wiki Principles“
KiWi - Knowledge in a Wiki
http://www.kiwi-project.eu
67
• content + semantic metadata (finished)
• transactions & versioning (mostly finished)
• semantic tagging (mostly finished)
• facetted search (in progress)
• social networking (in progress)
• personalisation (in progress)
• reasoning (in progress)
KiWi - Core Components
http://www.kiwi-project.eu
68
• KiWi Wiki (finished)
• TagIT (mostly finished)
• Dashboard (in progress)
• Blog (planned)
important:
content shared between applications!
KiWi - Applications
http://www.kiwi-project.eu
69
Demo!
http://showcase.kiwi-project.eu
70
Conclusion
71
• reimplementation on top of the KiWi platform
• integration of community features (social networking, sharing, ...)
• integration of different kinds of content(news, wiki, blogs, photos, ...)
• backed by advanced Semantic Web technology (reasoning, information extraction)
Where do we go?
72
Book tips ...
• Richard Saul Wurman: Information Anxiety 2
• David Weinberger: Everything is Miscellaneous
• Clay Shirky: Here Comes Everybody - the Powerof Organising without Organisatons
73
SNML Books (German)
Nachrichten 2.0: Eine Analyse internationaler Nachrichtenangebote im InternetISBN: 978-3-8370-5731-7
Erfolgreicher Aufbau von Online-Communitys: Konzepte, Szenarien und Handlungsempfehlungen (April 2009)ISBN: 978-3-902448-13-2
74
Thanks!
Dr. Sebastian Schaffert
| http://www.salzburgresearch.at| http://www.newmedialab.at
| http://www.kiwi-project.eu (KiWi Website)| http://planet.kiwi-project.eu (KiWi blog)
75