Loading…
smsociety14 has ended
View analytic
Saturday, September 27 • 16:51 - 17:10
"The Recommender Role of Twitter for Science Topics"

Sign up or log in to save this to your schedule and see who's attending!

Background: For most people, media are their main source of scientific information. Twitter offers a new channel to disseminate, consume, and debate research findings. The interface between the science community and the public is shifting from traditional news outlets and their websites to natively web-based services (Brossard & Scheufele 2013).

Objective: The paper aims to demonstrate how scientific topics are discussed on Twitter and in traditional news outlets. Who are the central players? What does the network structure tell us about the diffusion of knowledge? How are web resources and other users referenced in tweets? The overall motivation for this research was to investigate how social media may enhance the flow of research-based knowledge.

Methods: During five weeks, 965 online science articles form six traditional news outlets (selected based on a classification by Weber & Monge 2011) were automatically collected and analyzed. A continuous topic detection method of public scientific issues was realized using latent Dirichlet allocation (see Blei 2012). This enabled the generating of a corresponding 35 day sample of 72,469 Twitter messages, retrieved by searching the API with the science news topic keywords (e.g. “climate change”).
Social network analysis is used to examine structures in the Twitter users who authored tweets about scientific issues. Network connections are considered present if the individual was mentioned, replied to, or had a post retweeted. With regards to the textual contents of the tweets and articles, topic models, co-occurrence analyses (Grimmer & Stewart 2013), multidimensional scaling, and sentiment analysis are explored. The unique feature of this paper is the dynamic linkage between tweets and news—this avoids an ex-ante restriction on a specific issue and its search terms (e.g. nanotechnology).

Results: Network analysis and visualization reveals patterns of interaction that characterize the science news Twitter community as comprising a large component with roughly 40% of the users. The indegree distribution is extremely skewed and reveals a set of dominant actors in the domain of science news. Unsurprisingly, these are professional news organizations such as The New York Times. Network structures indicative of conversational exchange are rare. A large proportion of science tweets, particularly as compared to tweets in general, contain a web link or mention another user in some way. Sentiment analysis shows that Science news are on average more positive than tweets, which in turn experience much greater sentiment variation. The contextualization of science issues (term co-occurrences) differs for some topics while it is essentially the same for others. Major topics in the time frame of data collection include Mars explorations, the Nobel Prizes, the U.S. government shutdown, breast cancer, and climate change.

Conclusions: The results of the network and link analysis point to a recommender role of Twitter as the service moves from a social interaction medium to a global information network, as others have noted (e.g. Braun & Gillespie 2011; van Dijck 2011). The empirical results and the literature review insights are synthesized in a bigger picture to foster future research.

References:
Blei, D. M. (2012). Probabilistic topic models. Proceedings of the 17th ACM SIGKDD International Conference Tutorials (pp. 77–84).New York: ACM. http://dx.doi.org/10.1145/2107736.2107741

Braun, J., & Gillespie, T. (2011). Hosting the public discourse, hosting the public. Journalism Practice, 5(4), 383–398. http://dx.doi.org/10.1080/17512786.2011.557560

Brossard, D., & Scheufele, D. A. (2013). Science, new media, and the public. Science, 339(6115), 40–41. http://dx.doi.org/10.1126/science.1232329

Grimmer, J., & Stewart, B. M. (2013). Text as data: the promise and pitfalls of automatic content analysis methods for political texts. Political Analysis, 21(3), 267–297. http://dx.doi.org/10.1093/pan/mps028

Van Dijck, J. (2011). Tracing Twitter: the rise of a microblogging platform. International Journal of Media & Cultural Politics, 7(3), 333–348. http://dx.doi.org/10.1386/macp.7.3.333_1

Weber, M. S., & Monge, P. (2011). The flow of digital news in a network of sources, authorities, and hubs. Journal of Communication, 61(6), 1062–1081. http://dx.doi.org/10.1111/j.1460-2466.2011.01596.x

 


Speakers
avatar for Moritz David Büchi

Moritz David Büchi

University of Zurich


Saturday September 27, 2014 16:51 - 17:10
TRS 1-003 Ted Rogers School of Management

Attendees (10)