99486 Motivated by the significant decline in citizen’s trust in governments over the past decades, this paper explores how policy decision makers and researchers can use social media analytics to investigate “trust”, specifically the relationship among trust in government, trust in state institutions and citizens’ collective behavior. Analysis of these complex socio -political issues using online social data requires a human in the inference loop while also benefiting from computational methods to handle large amounts of unstructured data and the inference of relevant data features. This project implemented an exploratory framework from data collection to data analysis using, among other methods, a set of tools developed by project team members that integrate visualization, interactivity and machine learning. Visual analytics provided the team with knowledge of interactive methods and investigative methodological approaches while predictive modeling provided methods to operationalize difficult concepts such as “trust” through a proxy of sentiment classification of opinion in social media data. To highlight the power of a mixed-initiative visual analytics-data science approach, this technical note describes the exploratory analysis work undertaken for analysis of collections of Tweets from Brazil, and describes further work that conceives data science methods to assist the analysis process by supporting definition of constructs of concepts of interest using social media data, and assisting the evaluation of evidence for hypotheses evaluation in an interactive-machine learning fashion. The outcomes of this project aim to support social sciences inquiry using observational social media data and World Bank operations. CONTACT Victoria Lemieux, Senior Public Sector Specialist, Integrated Digital Solutions Group, vlemieux@worldbank.org CONTACT Victoria Lemieux, Senior Public Sector Specialist, Integrated Digital Solutions Group, vlemieux@worldbank.org Figure 1: Summary data for a collection of Twitter data with overlay of events and Sentiment Horizon Charts CONTACT Victoria Lemieux, Senior Public Sector Specialist, Integrated Digital Solutions Group, vlemieux@worldbank.org 1 OECD (2013) Government at a Glance 2013, OECD, Paris 2 Dhar, V. (2013). Data science and prediction. Communications of the ACM, 56(12), 64–73. doi:10.1145/2500499 3 Provost, F., & Fawcett, T. (2013). Data Science for Business. O’Reilly Media. 4 Klandermans, B., & Roggeband, C. (Eds.). (2007). Handbook of social movements across disciplines. Springer Science & Business Media. 5 Op Cit. 6 Ribarsky, W., Fisher, B., & Pottenger, W. M. (2009). Science of Analytical Reasoning. Information Visualization, 8(4), 254–262. doi:10.1057/ivs.2009.28 7 Edwards, M. (2015). The trust deficit - concepts and causes of low public trust in governments. Committee of Experts on Public Administration, 14th Session. E/CN.16/2015/CRP.3/Rev.1. 8 Thelwall, M., Buckley, K., Paltoglou, G. Cai, D., & Kappas, A. (2010). Sentiment strength detection in short informal text. Journal of the American Society for Information Science and Technology, 61(12), 2544–2558 9 Boertjes, E., Gerrits, B., Kooij, R., van Maanen, P. P., Raaijmakers, S., & de Wit, J. (2012). Towards a social media-based model of trust and its application. In ICT Critical Infrastructures and Society (pp. 250-263). 17 Wong PC, EG Hetzler, C Posse, MA Whiting, Springer Berlin Heidelberg. SL Havre, NO Cramer, A Shah, M Singhal, AE 10 Hetherington, M. J. (2005). Why trust matters: Turner, and J Thomas. (2004). "IN-SPIRE InfoVis Declining political trust and the demise of 2004 Contest Entry." In IEEE Symposium on American liberalism. Princeton University Press. Information Visualization. InfoViz 2004. 11 Newton, K., & Norris, P. (2000). Confidence 18 Heer, J., Kong, N., & Agrawala, M. (2009). in public institutions. Disaffected democracies. Sizing the horizon: the effects of chart size and What's troubling the trilateral countries layering on the graphical perception of time 12 Op Cit. series visualizations. In Proceedings of the 13 Kristensen, C.H., Gomes, C.F.A., Justo, A.R., SIGCHI Conference on Human Factors in Vieira, K. (2011). Brazilian norms for the Computing Systems (pp. 1303-1312). ACM Affective Norms for English Words. Trends 19 Arias-Hernandez, R., Kaastra, L. T., Green, T. Psychiatry Psychother. 33(3) – 135-146. M., & Fisher, B. (2011). Pair Analytics: 14 Hemsley, J., Ceskavich, B., Tanupabrungsun, Capturing Reasoning Processes in S. (2014). STACK (Version 1.0). Syracuse Collaborative Visual Analytics. In 2011 44th University, School of Information Studies. Hawaii International Conference on System Retrieved from Sciences (HICSS) (pp. 1–10). https://github.com/bitslabsyr/stack DOI: 20 Heuer, R. J. (1999). Psychology of 10.5281/zenodo.12388 intelligence analysis. Center for the Study of 15 Bamburic, M. (2012). Twitter: 500 million Intelligence, Central Intelligence Agency. accounts, billions of tweets, and less than one 21 Heuer, R. J. (1999). Psychology of percent use their location. Retrieved from intelligence analysis. Center for the Study of http://betanews.com/2012/07/31/twitter-500- Intelligence, Central Intelligence Agency. million-accounts-billions-of-tweets-and-less- 22 Folger R. (1986). Rethinking equity theory: A than-one-percent-use-their-location/ referent cognitions model. In Bierhoff HW, 16 Shulman, S. (2011). DiscoverText: Software Cohen RL and Greenberg J (eds) Justice in Training to Unlock the Power of Text. UMass Social Relations. Plenum, New York, 145–162. Amherst, Political Science Department. CONTACT Victoria Lemieux, Senior Public Sector Specialist, Integrated Digital Solutions Group, vlemieux@worldbank.org