micro-serendipity: meaningful coincidences in everyday life shared on twitter

28
Micro-serendipity: Meaningful Coincidences in Everyday Life Shared on Twitter iConference 2013, Fort Worth, TX Toine Bogers & Lennart Björneborn Royal School of Library and Information Science, Copenhagen

Upload: lennart-bjoerneborn

Post on 08-May-2015

1.586 views

Category:

Social Media


0 download

DESCRIPTION

Presentation at iConference 2013, Feb. 13, 2013, Fort Worth, Texas. Fulltext paper available: http://hdl.handle.net/2142/36052 Bogers, T. & Björneborn, L. (2013). Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter. Proceedings of iConference 2013, pp. 196-208.

TRANSCRIPT

Page 1: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

Micro-serendipity: Meaningful Coincidences

in Everyday Life Shared on Twitter

iConference 2013, Fort Worth, TX

Toine Bogers & Lennart Björneborn

Royal School of Library and Information Science, Copenhagen

Page 2: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

2

Page 3: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

motivation (1/3)

why is serendipity interesting?

serendipity: finding interesting things in unplanned ways

important role in many scientific discoveries

also integral part in everyday information behavior

how we get new inspiration, ideas, insights in everyday life

the very way we learn many new things in life since infanthood

design for stimulating and supporting serendipity

search engines, recommender systems (e.g., music), micro-

blogging, …

3

Page 4: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

motivation (2/3)

needed: better understanding

different definitions focus on different aspects:

include active (foreground) interest?

relate to latent (background) interest alone?

needed: better understanding how people experience and

communicate serendipitous occurrences in everyday life

naturalistic studies of everyday serendipity

based on data generated by users themselves (Erdelez, 2004)

most previous studies based on data elicited from interviews

everyday serendipitous experiences of bloggers (Rubin et al., 2011)

4

Page 5: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

micro-serendipity: investigating contexts and attributes of

everyday serendipity as shared on Twitter

we use non-elicited, self-motivated user data from Twitter

we omit a preset definition of serendipity

understand what users themselves consider as serendipitous

experiences and how they actually describe these experiences

Twitter: window into everyday life of millions of users

everyday experiences, interests, conversations, language use

5

motivation (3/3)

micro-serendipity on Twitter

Page 6: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

research questions

RQ 1 What types of serendipity do Twitter users

experience?

RQ 2 How often do people share serendipitous

experiences on Twitter?

RQ 3 What terminology do people use on Twitter to

describe their serendipitous experiences?

6

Page 7: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

crawled 30,000+ English-language tweets containing the

term ‘serendipity’ from Aug 2011–Feb 2012

used Topsy, social media search engine to access tweets

can search further back in time than Twitter

access to max. 1% of all tweets

no obvious crawling bias, so assumed to be representative

7

methodology (1/4)

data collection

Page 8: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

open coding approach to develop coding categories

on Feb 2012 tweets

category of interest: PERS (personal)

clearly describe personal insight or experience of a

serendipitous occurrence on the part of the tweeter

we tried to eliminate our pre-conceptions of what serendipity is

used context (included URLs and surrounding tweet stream)

to disambiguate

8

methodology (2/4)

coding tweets

Page 9: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

applied coding scheme to last three months of tweets

with the hashtag #serendipity (Dec 2011–Feb 2012)

open coding phase showed #serendipity more likely to contain

PERS tweets

inter-annotator agreement of 0.65

remaining differences resolved through discussion

coded 1073 tweets with 14.9% (N=160) in PERS category

9

methodology (3/4)

coding tweets

Page 10: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

10

methodology (4/4)

‘serendipity’ noise

Page 11: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

findings: RQ1 (1/4)

serendipity context: leisure vs. work

RQ 1 What types of serendipity do users experience?

qualitative analysis of 160 tweets in PERS category

distinction between leisure- and work-related activities

141 tweets (88.1%) leisure-related

14 tweets (8.8%) work-related

1 tweet coded as both; 4 tweets too ambiguous to code

rich diversity in leisure-related activities connected to

serendipitous experiences

all kinds of digital and physical spaces

including media, shopping, sports and transportation

11

Page 12: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

12

work- and

leisure-

related

Page 13: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

13

work-related

Page 14: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

14

leisure-

related

Page 15: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

unplanned

everyday incidents

unanticipated eureka

moments in science

different serendipity thresholds

when does a user find something unusual, unexpected, or surprising

enough to consider it as serendipity?

plain novelty or pleasant diversion may sometimes be enough

serendipity is a highly subjective phenomenon

serendipity continuum

different degrees of surprise:

serendipity is not a discrete concept

findings: RQ1 (2/4)

serendipity thresholds & continuum

15

Page 16: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

16

serendipity thresholds

Page 17: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

findings: RQ1 (3/4)

background + foreground serendipity

background serendipity (‘traditional’ serendipity)

unexpectedly finding something meaningful related to a background

interest; changing a person’s focus and direction

foreground serendipity (‘synchronicity’)

unexpectedly finding something meaningful related to a foreground

interest/preoccupation; confirming a person’s focus and direction

in everyday experiences and in science (e.g., Makri & Blandford, 2012)

both types of serendipity deal with people experiencing

meaningful coincidences

people considering an occurrence as both meaningful and incidental

17

Page 18: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

18

foreground serendipity (‘synchronicity’)

Page 19: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

findings: RQ1 (4/4)

key elements in serendipity

unexpectedness + insight + value (Makri & Blandford, 2012)

unexpectedness + value + preoccupation

some degree of insight always present in order to consider an

occurrence as both unexpected/incidental and valuable/meaningful;

– i.e., considering the occurrence as a meaningful coincidence

19

Page 20: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

20

unexpectedness + value + preoccupation

Page 21: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

21

unexpectedness + value + preoccupation

Page 22: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

findings: RQ2

frequency of sharing serendipity

RQ 2 How often do people share serendipitous

experiences on Twitter?

160 PERS tweets from 146 different users

tweets from all users with >1 PERS tweets were identical

repetitions

extended this to the full 7-month, 30,000+ tweet crawl

only a handful users had more than one tweet about serendipity

not that common a (re-)occurrence on Twitter!

we only focused on only one way of describing serendipity

22

Page 23: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

findings: RQ3 (1/3)

describing serendipity

RQ 3 What terminology do people use on Twitter to

describe their serendipitous experiences?

two reasons for answering this question

general interest in how people describe serendipitous occurrences

can we train an automatic classifier to pick out PERS tweets?

focused on three ways of signaling serendipity

words

part-of-speech tags (e.g., noun, past tense verb, …)

hashtags (e.g., #serendipitous, #insight, …)

used log-likelihood to extract representative signals

measures how surprising the usage of a signal between two text

collections is 23

Page 24: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

findings: RQ3 (2/3)

describing serendipity

words

PERS:

just, found, noticed, bumped, simultaneously, immediately, omg

non-PERS:

watching, serendipity, Kate, John, movie, chocolate, sundae

no conclusive identification of serendipity vocabulary

parts-of-speech

past tense verbs more often used in PERS tweets

present tense verbs more often used in non-PERS tweets

nouns more likely in non-PERS tweets

24

Page 25: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

findings: RQ3 (3/3)

describing serendipity

hashtags

hashtags most commonly co-occurring with #serendipity belong

to events: #nyc, #superbowl, #weezercruise, #saints

promising hashtags for future work:

#serendipitous, #synchronicity, #chance, #insight,

#randomness, #accident, #wtf, #lucky, #surprise

combination of different signals seems to show promise

in automatic classification of PERS tweets

25

Page 26: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

conclusions

RQ 1: no single type of serendipity

people experience this along a continuum with different thresholds

RQ 2: serendipity appears to be a rarely tweeted phenomenon

perhaps because it is uncommon or in fact too common?

longitudinal studies are necessary to confirm this though

RQ 3: no single signal singles out serendipitous occurrences

combination of different signals shows promise for automatic

classification

26

Page 27: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

future work

actual word usage on Twitter may suggest terms for other

serendipity studies

developing an automatic serendipity classifier

include data from surrounding tweets in tweet stream

investigate how people describe matches between

environmental factors and foreground/background interests

include differences between physical and digital environments

27

Page 28: Micro-serendipity: Meaningful coincidences in everyday life shared on Twitter

questions? comments?

28

Lennart Björneborn @connecto

Toine Bogers @toinebogers