Tutorials



Tutorial, March 28, 8:30am–12:30pm:

Toolkits for Computational Social Science: Using Honest Signals to Predict and Shape Human Responses
    Alex `Sandy' Pentland, MIT


Tutorial, March 28, 2:00–5:30pm:

Social Network Analysis of Personal and Group Networks
    Allen Tien, Medical Decision Logic
    Chris McCarty, University of Florida Survey Research Center
    Eric C Jones, University of North Carolina-Greensboro


Tutorial, March 28, 2:00–5:30pm:

Understanding Social Media: Tools, Applications, and Processes
    Nitin Agarwal, University of Arkansas at Little Rock


Toolkits for Computational Social Science: Using Honest Signals to Predict and Shape Human Responses

Alex `Sandy' Pentland, MIT
Tutorial Slides: funf Mobile Sensing System
Tutorial Slides: Influence Model

During the last decade we have developed measurement toolkits based on electronic badges, smart phones, and signal processing that allow us to accurately quantify human behavior in everyday situations on a continuous basis over long time periods. Using these toolkits we have developed a consistent mathematical modeling approach that accurately predicts outcomes of interactions at the scales of the individual, small group, organization, city, and nation-state. Equal error prediction accuracies for subject behaviors typically range from 80% to 95%, with 40 to 60 percent of the variance modeled. Applications to health outcomes and cultural outcomes have similar accuracies.

This suite of technologies has been called `Breakthrough idea of the year’ by Harvard Business Review, `a technology that will change the world’ by Technology Today, and Nature published a three-page report that described the work as `taking [modeling of human behavior] to a new level.’ Applications of these technologies were called `the next Google’ by Newsweek, were critical in winning of the 40th Anniversary of the Internet Grand Challenge, and have been central to the social computing strategies of both DARPA and the World Economic Forum members (multinational companies and government regulators).

The key insight is that humans today are not so different from humans in pre-history, before we had sophisticated language capabilities. Pre-linguistic social species coordinate by signaling, and in particular `honest signals' which in addition to conveying information reliably cause predictable responses by the listener. Examples of honest signaling include patterns of turn-taking, of small group interaction, and of association and communication. Experimental evidence suggests that modern language evolved `on top' of these ancient signaling mechanisms, and that today linguistic and signaling mechanisms operate in parallel, expressing different aspects of the same intentional states. Consequently, by measuring these `honest signals’ we can reliably predict outcomes.

In this tutorial I will describe the sociometric badges and Andriod platform sociometric software that we have developed, covering their function, capability, and typical use. These tools will be made available to interested participants. I will also cover the mathematical toolkit we have developed, describing the theory, capability, and typical use. These tools will also be made available to participants.

Finally, I will illustrate the use of our sociometric measurement tools together with our mathematical analysis tools on a variety of problems, including individual (passive screening for health problems), dyadic (assessing trustworthiness, screening for depression), small group (providing a real-time performance meter for groups), organizations (reengineering communication patterns for greater productivity), and large-scale sociocultural outcomes (diabetes risk, crime risk). For additional information see http://hd.media.mit.edu

About the instructor:

Prof. Alex `Sandy’ Pentland directs MIT’s Human Dynamics Laboratory and the MIT Media Lab Entrepreneurship Program, and advises the World Economic Forum, Nissan Motor Corporation, and a variety of start-up firms. He has previously helped create and direct MIT’s Media Laboratory, the Media Lab Asia laboratories at the Indian Institutes of Technology, and Strong Hospital’s Center for Future Health. Sandy is among the most-cited computational scientists in the world, and is a pioneer in computational social science, organizational engineering, mobile computing, image understanding, and modern biometrics. Profiles of Sandy have appeared in many publications, including the New York Times, Forbes, and Harvard Business Review. His most recent book is `Honest Signals,' published by MIT Press. See http://media.mit.edu/~pentland for additional detail.


Social Network Analysis of Personal and Group Networks

Allen Tien, Medical Decision Logic
Chris McCarty, University of Florida Survey Research Center
Eric C. Jones, University of North Carolina-Greensboro
Tutorial Slides

     Audience: This tutorial is designed for researchers, practitioners, program staff from federal agencies and graduate students interested in collecting, managing, analyzing and visualizing social network data in a user-friendly way.

Special Instruction: Participants should bring their own laptop.

Description: Social Network Analysis typically considers the connections between individuals, human behaviors, and/or human groups as well as the aggregates of these connections. However, people making use of the techniques available through social network analysis hail from many backgrounds, including virtually all social sciences, most physical sciences and many of the humanities. Application of social network analysis has been or is being attempted in all nearly aspects of business, education, development and other non-profit work, and government.
A main difficulty in engaging social network analysis are the technical aspects of data management and import/export for the various free and paid software available. This workshop is intended to increase accessibility of social network analysis, particularly the utility of visualizing network data for further exploring aspects of any research or applied question.
The tutorial will present concepts and demonstrates tools for personal network analysis and whole network analysis, primarily EgoNet and VisuAlyzer that run on Windows platforms. Relatively unique to EgoNet is the ability to author a survey and conduct that survey from the same piece of software that also visualizes the data, provides analytical capabilities, and exports datasets and visualizations. Also, hundreds of networks can be analyzed at once.
In subgroups at the workshop, participants will collect, visualize and analyze personal networks (e.g., interactions between people in your life), N x N whole networks (e.g., interactions among a list of people), and N x M whole networks (e.g., interactions between a list of participants with a list of the topics of research in which they are involved). This will include the opportunity to assess one’s own personal-professional network and explore linking that with the networks of other participants.

About the instructors:

Eric C. Jones is a research scientist at the University of North Carolina-Greensboro and has focused his recent work on understanding how the structuring of social relations following extreme events (e.g., natural disasters, pioneer colonization, and immigration) impacts individual outcomes and recovery—often through the use of Social Network Analysis. His general research interests concern cross-cultural vulnerability and resilience under conditions of environmental and/or social change. He recently coauthored the edited volume The Political Economy of Hazards and Disasters (2009, AltaMira Press), published the innovative methodological piece ‘‘Extreme Events, Tipping Points and Vulnerability: Methods in the Political Economy of Environment’’ in Environmental Social Sciences: Methods and Research Design (in press, 2010, Cambridge University Press), and performed the application of social network analysis to cooperation among pioneer colonists in ‘‘Wealth-Based Trust and the Development of Collective Action’’ (2004) in the journal World Development. Jones’ recent research has involved association of ego-centered network structure and content with disaster outcomes in Ecuador, Mexico and the US.

Allen Y. Tien is Founder, President, and Director of Applied Research of Medical Decision Logic. As well as being a Board-Certified psychiatrist, he has Master's level training in biostatistics and postdoctoral training in psychiatric epidemiology, research design and methods, and public mental health issues. Dr. Tien was a full-time tenure-track faculty member in the Johns Hopkins University School of Public Health Department of Mental Hygiene and School of Medicine Department of Psychiatry from 1988 to 1997, promoted to Associate Professor in 1996. Active in research in public mental health epidemiology, services, and prevention, and clinical neuroscience, for a number of years he taught a key Mental Hygiene course on multi-level etiologic models of mental disorders. He has been developing both software and hardware for computer-based assessment systems since 1987, and is knowledgeable about information technology design and engineering. He conceived of and continues to lead a number of Small Business Innovation Research (SBIR) projects that address psychometrics and the assessment of symptoms, diagnostic and treatment logic, indexed audio and video recording and playback, data management, graph (e.g., social networks, genetic networks) visualization, software tools for social network analysis (SNA), multi-level integration of genetic and phenotype data, nursing workflow, patient education, primary care behavioral health screening, and patient reported outcomes for palliative and hospice care.

Christopher McCarty is the Survey Director of the University of Florida Survey Research Center (UFSRC), a 70 station CATI lab that specializes in large-scale health-related surveys. The UFSRC is among the largest university-based surveys in the country and conducts surveys for the state of Florida and for other states. McCarty has also served as a consultant on survey topics in the Republic of Ghana, the Republic of Mali, and Mexico. Over the past 20 years McCarty has had an active research agenda in the area of social networks, specializing in ego-centered network research. These include the Reverse Small World, the Network Scale-up Method and extensive research into ego-centered network elicitation techniques. McCarty’s most recent interests have focused on structure in ego-centered networks and how that relates to behavioral outcomes. He conducted a study of depression and egocentric networks using a prototype version of EgoNet written in Java. He was the lead consultant in the SocioMetrica project that created the Windows version of EgoNet, and used it to study acculturation of migrants in Spain.


Understanding Social Media: Tools, Applications, and Processes

Nitin Agarwal, University of Arkansas at Little Rock
Tutorial Slides

Social media provide an inexpensive, easy-to-use, and almost ubiquitous platform as well as a dynamic, collaborative, democratic and un-regulated environment for Internet users to voice opinions, express beliefs, share thoughts, and participate in discussions.

Its open standards and low barrier to publication have transformed information consumers to producers. This has created a plethora of open-source intelligence, or "collective wisdom" that acts as the storehouse of overwhelming amounts of knowledge about the members, their environment and the symbiosis between them. Nonetheless, vast amounts of this knowledge still remain to be discovered and exploited in its most suitable way. Further this enables to advance our understanding of the induced individual and collective socio-technical behavioral changes creating new synergies and capabilities between social science and computational science as manifested by the emergent phenomenon of "unorganized organizations" exemplified by the applications ranging from the recent socio-political movements at transnational scales and diversity (such as, the recent Tunisian and Egyptian uprisings) to crisis management and disaster recovery.

The objective of this tutorial is to give a comprehensive overview of the techniques, applications, and research issues in the social media. Techniques for data collection, knowledge extraction (both content and graph-theoretic approaches), research methodologies, and model evaluation will be discussed highlighting challenges and opportunities. Specifically, through case studies it will be demonstrated that how concepts such as modeling, clustering and community extraction, influence, opinion mining, trust and privacy can offer insights into various organic processes observed in social networks.

Audience:

The intended audience for this tutorial mainly includes researchers, graduate students, and professionals who are new to this area or who have some acquaintance with Web 2.0 technologies and various social networking services. The audience is expected to have basic understanding of Web applications, search, and statistics. In addition, this tutorial also aims to encourage collaborative research in social media via fields like computer science, sociology, and anthropology among others.

About the instructor:

Nitin Agarwal is an assistant professor of Information Science at University of Arkansas at Little Rock. He holds a Ph.D. in Computer Science from Arizona State University with outstanding dissertation recognition. His research interests include social computing and behavioral modeling, data mining, knowledge extraction in social media, modeling and evaluation of influence, trust, homophile, and collective intelligence. This expertise is demonstrated through numerous highly cited articles in leading journals and participation in prestigious conferences (including a Best Paper Award in the IEEE International Conference on Privacy, Security, Risk, and Trust 2010), and through the publication of various books on Modeling and Data Mining in Blogosphere published by Morgan & Claypool and on Social Computing in Blogosphere: Challenges, Methodologies, and Opportunities published by LAP Lambert Academic Publishing AG & Co. KG. He has guest edited special issues on Social Computing in Blogosphere for IEEE Internet Computing magazine that appeared in March-April 2010 and Social Computational Systems for the Journal of Computational Science to appear in 2011. He has delivered well-received talks and tutorials on Blogosphere: Research Issues, Applications, and Tools (KDD2008), Community Detection and Behavior Study for Social Computing (SocialCom09), and on Identifying the Influential Bloggers in a Community (WSDM 2008). The videos of these talks can be found at http://videolectures.net/nitin_agarwal/ . He currently serves on program and technical committees of several prestigious conferences and Editor-in-chief for the Social Comp 2011 (http://social-comp.org/). His research is supported by the US Office of Naval Research. He has developed new courses in the area of social computing and incorporates various social media technologies in this pedagogical venture. More details can be found at: http://ualr.edu/nxagarwal/.