Tweets and Scientific Conferences: The Use Case of the Science 2.0 Conference

Tweets and Scientific Conferences

The Use Case of the Science 2.0 Conference

Athanasios Mazarakis and Isabella Peters

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 2

Agenda

2 Methods for Evaluation and Analysis

4 Conclusion

3 Results

5 Future Work

Introduction, Motivation and Research Questions1

• Time stamp for each tweet allows time-specific analysis• Heavy tweeting can be linked to the conference program• Previous studies often analyzed only on a daily basis

(Ross et al., 2011)

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 3

Introduction

• Rapid growing and best example: Twitter• Twitter as center of attention in scientific community• Use social network services for awareness support

Microblogging and Scientific Conferences

• „#“ – hashtag, to group and easily follow topics over time• „@“ – at, to directly reference and raise attention

Twitter signs

Timeline

(Java et al., 2007)

(Reinhardt et al., 2009)

(Ross et al., 2011)

(Reinhardt et al., 2011)

(Sherr, 2014)

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 4

Motivation

• Promote the conference and usage of a hashtag• General information and related aspects (dates, keynotes, …)• Increase excitement and establish a community

Before a conference

• Last minutes changes or announcements• Discussions on conference presentations

During a conference

• Thank attendees• Asking for feedback• References to related blogs, newspapers, …

After a conference

(Reinhardt et al., 2009)

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 5

Research questions

Are there any particular events or occasions that lead to peaks in Twitter activity during a scientific conference?1

Is it possible for the content-based analysis of the tweets (categorization and prediction) to develop a general codebook for scientific conferences?

• Purpose of tweet• Target of web link• Content itself

2

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 6

Agenda

2 Methods for Evaluation and Analysis

4 Conclusion

3 Results

5 Future Work

Introduction, Motivation and Research Questions1

•Tweet observation period: 14.03.2014 - 14.04.2014•Testbed: Science 2.0 conference in Hamburg, Germany

• 26th and 27th March, 2014; English speaking conference + live stream• 153 registered attendees (analysis for attendees and non-attendees)

•Hashtag for analysis: #sci20conf •Tweets from conference organizers: @lfvscience20•1879 tweets collected – 676 for the final analysis

• Excluded: 665 redundancy; 489 retweets, 49 modified tweets• 582 from attendees (86%), 94 from non-attendees (14%)

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 7

Methods for Evaluation and Analysis

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 8

Agenda

2 Methods for Evaluation and Analysis

4 Conclusion

3 Results

5 Future Work

Introduction, Motivation and Research Questions1

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 9

Results

Results comparable to other studies! (Ross et al., 2011)

• Mean: 6.83• SD: 15.62• Maximum: 95• Median: 1• 51.5% of users sent only one

tweet

• 1214 tweets• 489 retweets (40%)• 718 tweets with “@” sign (59%)

• 229 without retweets (19%)• 648 with link (53%)

• 287 without retweets (24%)

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 10

Results – peak analysis

Results counter results of other studies! (Ebner et al., 2010)

Support for the first research question!

• 822 tweets from conference attendees (68%) – 392 non-attendees (32%)• BUT: 200 retweets conference attendees (24%) – 289 non-attendees (74%)

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 11

Results – content analysis

• I = conference content• T = sharing resources• N = note• O = organization and

announcements• B = conditions of

conference• A = other

• Inter-rater reliability:• Fleiss‘ Kappa = .60 • Moderate• Agreement rate: 68%• All results p < 0.01

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 12

Results – content analysis – example

“Schoepflin: scholars can exploit the potential of the Internet only if everything is #OA”

“free wifi #sci20conf , choose Riverside Hotel, kostenlos, AGB and its a go!”

“Lightning Talks. #impressions #sci20confhttp://t.co/aGp8jVycIw”

“The list of participants will not be published online. #sci20conf”

“On my way to #sci20conf”

“Today's conference hashtags to follow: #dhd2014 (continued), #sci20conf, #c4l14”

Conference content

Conditions of the conference

Sharing resources

Organization andannouncements

Note

Other

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 13

Results – content analysis

• BI = pictures• KO = conference website• AR = article• WE = web portal• VI = video• FO = presentation slides• BL = pictures• DO = documents• OR = organization

• Inter-rater reliability:• Fleiss‘ Kappa = .85 • Substantial• Agreement rate: 88%• All results p < 0.01

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 14

Results – content analysis

• WIS = scientific methods• PRO = projects• SOC = social web• OPE = open science• SON = other• BIG = big data• KON = conferences• BIB = libraries• UNT = enterprises• BEG = terms analysis

• Inter-rater reliability:• Fleiss‘ Kappa = .58 • Moderate• Agreement rate: 48%• All results p < 0.05

Support for the second research question!

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 15

Additional peak analysis

32% more conference content during peaks! (63% vs. 43%)

No significant results for the URL class.

Content of tweets varied very much for the different peaks, in accordance

to the concurrent presentation.

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 16

Agenda

2 Methods for Evaluation and Analysis

4 Conclusion

3 Results

5 Future Work

Introduction, Motivation and Research Questions1

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 17

Conclusion

• 4 week analysis of Science 2.0 conference

• No noticeable activity before of after the conference• No equal distribution of tweets over time

• Codebook validation successful

• Twitter activity highest after lunch and coffee breaks• Any information or regulation of the conference

should happen at these moments!

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 18

Agenda

2 Methods for Evaluation and Analysis

4 Conclusion

3 Results

5 Future Work

Introduction, Motivation and Research Questions1

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 19

Future Work

• Sophisticated quantitative and qualitativein-depth analysis

• Factor analysis for hashtags and classes

• Comparison with other scientific conferences

• Comparison with Science 2.0 2015 conference• Preliminary result: codebook performed excellent

• Development of algorithms e.g. to stir up debates during less interesting presentations

Thank you very much for your attention!

Questions?

Contact: a.mazarakis@zbw.eu

Athanasios Mazarakis and Isabella Peters – ECSM 2015, 10 th of July, Porto, Portugal 20

Thank you!