Making Sense of Microposts (#Microposts2016) Computational Social Science Track Katrin Weller∗ Aba-Sah Dadzie∗ Danica Radovanović∗ GESIS Leibniz Institute for the Knowledge Media Institute, Basic Internet Foundation / Social Sciences The Open University UNIK Cologne, Germany Milton Keynes, UK Oslo, Norway katrin.weller@gesis.org aba-sah.dadzie@open.ac.uk danica@basicinternet.org ABSTRACT In academia there is increasing tendency toward interdis- For the second time, the #Microposts workshop features ciplinary work between computer science and other technical a track to highlight social science perspectives on micro sciences [11, 21] and the social sciences [8], increasingly re- communication structures in online environments. This pa- ferred to as Computational Social Science. Different social per introduces the #Microposts2016 (Computational) Social media tools provide an expressive medium for sharing with Science Track, which all contribute to connecting research others – both acquaintances and the general public – feel- methods and approaches in computer science and social sci- ings, needs, current status, or simple statements [16]. This ence. By providing a forum for closer interaction between provides solid ground for forming phatic expressions, which the two fields, the track is becoming a platform for interdisci- we also refer to as Microposts. A Micropost constitutes a plinary projects and new ideas to combine different method- small, brief message, theme or a single thought, quick and ologies and theories. For this year’s special track we see the easy to publish, and that, posted from a variety of plat- trend of relating Microposts to external demographics or forms and by very large numbers of individuals with as many survey data as a way to better understand Microposts in viewpoints and interests, collectively provide a rich source their broader contexts. of information and opinion about a range of topics. The first micro-sized social media posts were exchanged using text; while the term Micropost has evolved with ad- CCS Concepts vances in technology the term now rarely needs definition. •Information systems → World Wide Web; •Applied Text is still a popular means of expression, e.g., in a tweet, computing → Enterprise computing; status update or a comment in a news feed. The term is how- ever now used even outside the workshop to refer not just to brief text posts but as a means of sharing also other multi- Keywords media information – in photos, streaming and pre-recorded Microposts, Social Science, Web Science, Computational So- video and audio, with the caveat that the post remains small. cial Science, Internet science, Internet research, social me- Popular options include as an Instagram photo (with a hash- dia, user-generated content, online communication tag), a Snapchat quick video message or a three second In- stagram Boomerang video. 1. BACKGROUND The brevity of Microposts makes them a convenient, low- cost means for sharing information in the moment and on Accelerated development of Information-Communication the go, from a variety of personal devices, from the myriad Technologies (ICT) has a profound impact on socio-cultural new apps and portals built each year. The utility of Mi- relationships and processes; this presents a challenge for re- croposts is seen in that where direct access to the Internet searchers from multiple disciplines and backgrounds. In an is not available, for instance in remote, off-grid or under- interconnected world of information, different forms of com- served areas, apps exist to allow the next most convenient munication and social dynamics are formed, referring to the communication means, including mobile networks and even socio-technological processes that take place online. New radio, to be used to transport Microposts. On the other technologies shape information, communication and collab- hand, where traditional communication channels and media oration dynamics in different environments while contribut- outlets are overwhelmed or cut off, typically during disasters ing to persistent interdisciplinarity. and emergencies, or during uprisings and mass demonstra- ∗ tions, social media access via Internet access may provide All authors made equal contributions a lifeline or “connector” to emergency and information ser- Permission to make digital or hard copies of all or part of this work for personal or vices, family and friends [4, 10, 14]. The Ushahidi platform, classroom use is granted without fee provided that copies are not made or distributed for instance, has been used for crowdsourced translation and for profit or commercial advantage and that copies bear this notice and the full cita- relay of information in a number of crises, using SMS with tion on the firstc page. Copyright 2016Copyrights held byfor components of this work author(s)/owner(s); owned by copying others than permitted social media, notable particularly following the 2010 Haiti ACM only must be honored. for private andAbstracting academic with credit is permitted. To copy otherwise, or re- purposes. publish, Publishedto post ason servers part or to#Microposts2016 of the redistribute to lists, requires prior specific Workshop permission proceedings, earthquake. LINE, one of the most popular social media and/or a fee.online available Requestaspermissions from permissions@acm.org. CEUR Vol-1691 (http://ceur-ws.org/Vol-1691) tools in Japan, was born in the wake of the 2011 earthquake #Microposts2016, Apr 11th, 2016, Montréal, Canada. to aid communication [9, 12]. c 2016 ACM. ISBN 978-1-4503-2138-9. DOI: 10.1145/1235 · #Microposts2016 · 6th Workshop on Making Sense of Microposts · @WWW2016 2. TOWARD MORE INTERDISCIPLINARY One modification was to the track name, to (Computational) APPROACHES FOR MAKING SENSE OF Social Science. The new name better reflects the intent to in- clude contributions from the social sciences as additional dis- MICROPOSTS ciplines on the one hand, and also contributions that clearly The #Microposts workshop [1] aims to provide an in- apply multidisciplinary approaches and bridge the social sci- terdisciplinary forum for Computer Science researchers and ences and computer science on the other. Submissions ac- practitioners to collaborate with specialists from other fields, cepted in 2015 demonstrated very well how cross-boundary including Information, Web, Social and Computational Sci- submissions enrich the overall workshop scope, and those in ences, to discuss the generation and analysis of Micropost 2016 reinforced the value in such work to making sense of data and promote effective application of its knowledge con- Microposts. tent in different contexts and situations, including emer- gency response, crowd and event tracking, mass communi- 3.1 Track Sponsor: GESIS cation and marketing, opinion mining and sentiment analy- For the second time GESIS, the Leibniz Institute for the sis. The track also looks at research examining ethics, legal Social Sciences [2] is our sponsor for this special track. As a and privacy issues in the analysis and reuse of data which, research infrastructural organisation for the social sciences, although typically published in public or semi-public fora, GESIS offers support and services for different phases in the often includes sensitive, identifiable information about indi- research process. The established services for supporting viduals. The social and communication dynamics that result traditional social science research have recently been com- from the use of Micropost-based services are sometimes im- plemented with a GESIS department in Computational So- pacted by cultural, generational and regional differences [9], cial Science, which focuses on algorithms and theories for often seen in data journalism and citizen reporting of civil studying social phenomena based on Web data. This work action and politics as in the use of Microposts in the Arab is in many ways closely related to research on making sense Spring, and in the use of the forum for education and citizen of Microposts. Support from GESIS confirms the relevance empowerment. of the special track; by helping to raise the profile of the Making a significant contribution to today’s “big data”, track, the Microposts workshop continues to attract partici- the research and analytical activity necessary in making pation from authors and attendees who would not normally sense of Microposts rely in large part on techniques and tools participate in a conference targeted predominantly at Com- for large-scale information extraction and knowledge acqui- puter Science. sition, in order to handle the very high rate at which Mi- croposts are published, and increasingly using multi-media. 3.2 Topics of Interest However, automated analysis alone cannot decipher the sub- language necessary to capture complete messages in such In 2016 three submissions were received for the (Computa- small packets, or the nuance often used to aid encoding of tional) Social Science track. A fourth submission submitted these very small snippets of information, and in public or to the main track was also included in the review process for semi-public fora, load message content. Key to unpack- this track as it clearly crossed interdisciplinary boundaries, ing Micropost content is the human in the analysis loop, with two reviewers from each track’s committee assigned to to identify and interpret nuance that cannot be picked up it. Out of these submissions, one was accepted for full paper by automated methods, to better understand why end users presentation and two as poster presentations. employ this medium and in what circumstances it may be All three accepted submissions contribute valuable per- seen as a preferred means of communication. In addition to spectives to the understanding of Microposts, using differ- large-scale computational analysis it may also be necessary ent disciplinary approaches to bridge the gap between the- to look more closely at selected cases – and combine this with ory and practice. However, after a second review round a other quantitative and, importantly, qualitative approaches decision was made not to award the best paper prize (Com- for analysis. It is also often necessary to combine different putational) Social Science track, as the standard for doing sources of data which complement each other. Further, from so was not fully reached. Honourable mention however goes a social science perspective, approaches that help to relate to the paper Comparing Social Media and Traditional Sur- online communities to wider online and offline populations veys Around the Boston Marathon Bombing [6]. The au- are particularly interesting. thors, Cody Buntain, Erin McGrath, Jennifer Golbeck and Gary LaFree, address one of the key challenges in bridging traditional social science methods with new computational 3. THE #MICROPOSTS2016 (COMPUTA- approaches. In a case study about the bombings during the Boston Marathon in April 2013 Buntain et al., combine sur- TIONAL) SOCIAL SCIENCE TRACK vey data (from a panel survey and an experimental setup) While the #Microposts workshops have always sought to with digital trace data from social media. They thus placed bring together researchers with different disciplinary back- Microposts, on Twitter, in the broader context of opinions grounds, the need to place an even stronger focus on contri- examined through survey data. Of value especially to com- butions from the broader field of social sciences became ap- puter scientists is how this paper shows capability to gain parent. #Microposts2015 thus featured the first dedicated insight into the value of survey data, as a way to enrich Social Science track in the #Microposts series, and the pro- online communication data. Social scientists on the other gramme committee was broadened to assemble even more hand will find information that helps them to judge the costs social scientists and researchers from related disciplines en- (temporal and financial) of collecting new types of data in gaged in the study of Microposts and online communication. comparison to standard approaches employed in the field. Based on the success in its first year, which saw two accepted As the authors conclude, applied together, survey and so- submissions, the Social Science track was continued in 2016. cial media data can show a more complete picture of public 30 · #Microposts2016 · 6th Workshop on Making Sense of Microposts · @WWW2016 opinion. the role of Micropost messages in dealing with unexpected Both poster presentations in this track present studies on events is another popular topic, having seen practical use in a specific group of Twitter users, and in both cases refer to crises, environmental disasters and emergencies. For exam- these users as some sort of elite. The poster Studying the ple, #Microposts2015’s Social Science track also featured a Role of Elites in U.S. Political Twitter Debates by Sebas- case study that investigated communication around a cri- tian Stier [20] concentrates on “elites” that dominate polit- sis event – the 2015 Charlie Hebdo Shooting in France [7]. ical conversations: politicians, other government represen- The 2016 paper on the Boston Marathon bombing [6] does tatives and news media representatives. Alex Jeongwoo Oh not delve as deeply into understanding the polarisation of & Pramuan Bunkanwanicha in CEOs on Twitter [13] focus different groups reacting to the terrorist attacks; it rather on “elite” users in different business sectors. They analyse reminds us that Microposts can be further enriched with the tweets of executives of major firms to detect differences additional data. As we see in this workshop, from a social in style and practice and relate them to demographics on science perspective, this may be opinions polled via surveys the one hand and economic performance of the respective as well as demographics collected for specific user groups. companies on the other. Studying Microposts based on selected groups of users 3.3 Track Committee instead of specific topics (e.g. as expressed through hash- The proposal to include again the (Computational) Social tags) enables different forms for understanding the ongoing Science track was strengthened by the programme commit- communication. Studying “elite” users such as politicians tee, who work in the Social Sciences, Computer Science and and CEOs has another direct advantage: once an account in Business Administration in Higher Education in Europe, is verified as belonging to a certain elite user it typically the Middle East and North and South America. Together, also becomes possible to work with additional information our committee comprise a wealth of research that was ex- about this person, such as demographics like age and loca- hibited in the the informative reviews for all submissions, tion. Even more detailed biographical information is usually whether accepted or not. This feedback resulted in final available from verified external sources for such elites. For papers that should provide a good contribution to the liter- instance, information about politicians may include political ature on Making Sense of Microposts and the broader fields programmes, information on CEOs may include wages and of related research. different positions held during their career. On the other hand, when studying regular users (non-public figures) one is usually bound to the information these users provide on Acknowledgments public (Web) platforms when filling in user profiles. Such Katrin Weller is senior researcher and team leader at the De- profile information is often limited, incomplete, ambiguous partment of Computational Social Sciences, GESIS Leibniz or even deliberately misleading. On Twitter, for example, Institute for the Social Sciences in Cologne, Germany. Aba- user profiles do not include information on gender or age Sah Dadzie is researcher in Data Science at KMi, the Open and there is no standardised form for entering geo-location. University, working on the EU project EDSA (no. 643937). This has inspired research on methods to automatically in- Danica Radovanović is a senior visiting researcher at UNIK, fer demographic information or other user attributes from Kjeller, Norway and a Digital Equality Advisor at the Basic online activities or user networks [5]. This is not without Internet Foundation headquartered in Oslo, Norway. its challenges, also because there is still ongoing debate in the Internet research ethics community about how much in- formation may be collected or inferred about regular Web 4. REFERENCES users, who may not have explicitly given “informed consent” [1] #Microposts2016 website. on how and if their data may be used for research or other http://microposts2016.seas.upenn.edu. purposes [3, 22, 23]. Focusing on elite users and their pub- [2] GESIS. http://www.gesis.org. lic roles through official accounts on Web platforms reduces [3] S. Alim. An initial exploration of ethical research much of the ethical concerns in social media research, as it practices regarding automated data extraction from becomes possible to work with publicly available, official in- online social media user profiles. First Monday, 19(7), formation from external, verified sources. Further, Microp- 2014. osts from such accounts are usually official statements rather [4] N. Antoniou and M. Ciaramicoli. Social media in than private statements or opinion. This has enabled, there- disaster cycle – useful tools or mass distraction? In fore, publication of some of the underlying research data 64th International Astronautical Congress, 2013. [19] utilised in the study of political elites by Stier [20] in [5] V. Beretta, D. Maccagnola, T. Cribbin, and #Microposts2016. This serves to provide both a valuable E. Messina. An interactive method for inferring contribution to other research on this and related topics, demographic attributes in twitter. In Proc., 26th ACM and is in line with public policy toward increasingly open Conference on Hypertext & Social Media (HT ’15), data, to enrich research and lead to improvements in use of pages 113–122, 2015. the collective intelligence gathered through media such as [6] C. Buntain, E. McGrath, J. Golbeck, and G. LaFree. Microposts. Comparing social media and traditional surveys We observe that some topics from the 2015 workshop were around the boston marathon bombing. In picked up again but from new perspectives, applying new Preoţiuc-Pietro et al. [15], pages 34–41. foci and methods. Political communication clearly is a per- [7] F. Giglietto and Y. Lee. To Be or Not to Be Charlie: sistently popular topic in this field (though the focus on elite Twitter hashtags as a discourse and counter-discourse users as in the 2016 case study is possibly a niche area). in the aftermath of the 2015 Charlie Hebdo shooting Dealing with crises through social media and understanding in France. In Rowe et al. [18], pages 33–37. 31 · #Microposts2016 · 6th Workshop on Making Sense of Microposts · @WWW2016 [8] R. H. Heiberger and J. R. Riebling. Installing [16] D. Radovanović and M. Ragnedda. Small talk in the computational social science: Facing the challenges of digital age: Making sense of phatic posts. In Rowe new information and communication technologies in et al. [17], pages 10–13. social science. Methodological Innovations, 9, 2016. [17] M. Rowe, M. Stankovic, and A.-S. Dadzie, editors. [9] F. E. Jandt. Dimensions of culture. In An Introduction Proceedings, 2nd Workshop on Making Sense of to Intercultural Communication Identities in a Global Microposts (#MSM2012): Big things come in small Community, pages 160–190. SAGE Publications, Inc, packages, Lyon, France, 16 April 2012, April 2012. 8th edition, 2016. [18] M. Rowe, M. Stankovic, and A.-S. Dadzie, editors. [10] J.-Y. Jung. Social media use and goals after the great Proceedings, 5th Workshop on Making Sense of east japan earthquake. First Monday, 17(8), 2012. Microposts (#Microposts2015): Big things come in [11] D. Lazer, A. Pentland, L. Adamic, S. Aral, A.-L. small packages, Florence, Italy, 18th of May 2015, Barabási, D. Brewer, N. Christakis, N. Contractor, May 2015. J. Fowler, M. Gutmann, T. Jebara, G. King, M. Macy, [19] S. Stier. Elite actors in the U.S. political D. Roy, and M. Van Alstyne. Computational social Twittersphere. GESIS datorium. science. Science, 323(5915):721–723, 2009. http://dx.doi.org/10.7802/1178, 2016. [12] H. McCracken. How Japan’s LINE app became a [20] S. Stier. Studying the role of elites in U.S. political culture-changing, revenue-generating phenomenon. twitter debates. In Preoţiuc-Pietro et al. [15], pages Fast Company Magazine, Mar 2015. 43–45. [13] A. J. Oh and P. Bunkanwanicha. CEOs on Twitter. In [21] M. Strohmaier and C. Wagner. Computational social Preoţiuc-Pietro et al. [15], pages 46–48. science for the world wide web. IEEE Intelligent [14] B. D. M. Peary, R. Shaw, and Y. Takeuchi. Utilization Systems, 29(5):84–88, Sept 2014. of social media in the East Japan earthquake and [22] K. Weller and K. E. Kinder-Kurlanda. I love thinking tsunami and its effectiveness. Journal of Natural about ethics: Perspectives on ethics in social media Disaster Science, 34(1):3–18, 2012. research. In Selected Papers of Internet Research [15] D. Preoţiuc-Pietro, D. Radovanović, A. E. (SPIR), Proc., ir15 – Boundaries and Intersections, Cano-Basave, K. Weller, and A.-S. Dadzie, editors. 2014. Proceedings, 6th Workshop on Making Sense of [23] M. Zimmer. “but the data is already public”: on the Microposts (#Microposts2016): Big things come in ethics of research in facebook. Ethics and Information small packages, Montréal, Canada 11th of Apr 2016, Technology, 12(4):313–325, 2010. 2016. 32 · #Microposts2016 · 6th Workshop on Making Sense of Microposts · @WWW2016