Computational Social Science Institute

University of Massachusetts Amherst

Lecture series generously sponsored by Yahoo!

Videos of some CSSI seminars are available  here

Fall 2014

Faculty Convocation
Friday, September 12, 2014 • 11:00AM
Bowker Auditorium, Stockbridge Hall

During the ceremony, Provost Katherine Newman will present the keynote address and eight nationally acclaimed faculty members will be presented with the Award for Outstanding Accomplishments in Research and Creative Activity.

Michael Ash, Economics; Center for Public Policy and Administration
W. Bruce Croft, Computer Science
David R. Evans, Educational Policy, Research, and Administration
Lyn Frazier, Linguistics
Panayotis Kevrekidis, Mathematics and Statistics
Barbara Krauthamer, History
Young Min Moon, Art, Architecture and Art History
Shelly Peyton, Chemical Engineering

For more information, contact events@umass.edu or phone 413-577-1101.

Peter Dodds

Peter Dodds

Professor, University of Vermont
Measuring Happiness, Health, and Social Stories, the Big Data Way
Friday, September 19, 2014 • 12:00PM-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  In this talk, I will report on a wide array of findings obtained through our real-time, remote-sensing, non-invasive, text-based `hedonometer'---an instrument for measuring positivity in written expression, soon to be housed online at hedonometer.org. I'll show how we have improved our methods to allow us to explore collective, dynamical patterns of happiness found in massive text corpora including the global social network Twitter, song lyrics, blogs, political speeches, and news sources. From the viewpoint of Twitter, I will report on global levels of temporal, spatial, demographic, and social variations in happiness and information levels, as well as evidence of emotional synchrony and contagion. I will also discuss how natural language appears to contain a striking frequency-independent positive bias, how this phenomenon plays a key role in our instrument's performance, and its connections with collective cooperation and evolution.

Bio:  Peter Sheridan Dodds is a Professor at the University of Vermont (UVM) working on system-level problems in many fields, ranging from sociology to physics. He is Director of the UVM's Complex Systems Center, co-Director of UVM's Computational Story Lab, and a visiting faculty fellow at the Vermont Advanced Computing Core. He maintains general research and teaching interests in complex systems and networks with a current focus on sociotechnical and psychological phenomena including collective emotional states, contagion, and stories. His methods encompass large-scale sociotechnical experiments, large-scale data collection and analysis, and the formulation, analysis, and simulation of theoretical models. Dodds's training is in theoretical physics, mathematics, and electrical engineering with formal postdoctoral and research experience in the social sciences. Dodds is currently funded by an NSF CAREER grant awarded by the Social and Economic Sciences Directorate.

Rodrigo Zamith

Rodrigo Zamith

Doctoral candidate, School of Journalism and Mass Communication, University of Minnesota
http://www.rodrigozamith.com/
What Computational Social Science Means for Traditional Modes of Media Analysis
Friday, October 3, 2014 • 12:00PM-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  The abundance of digitized data has become a defining feature of modern life, and particularly of modern communication as it is expressed through digital, social, and mobile platforms. For communication and media research, in particular, the possibilities of understanding communicative practices, social behavior, and the diffusion of information are great. However, this abundance also brings with it a number of challenges for media scholars as they struggle to deal with ever-larger volumes of data and seek out computational solutions. In this talk, I focus on the following question: What does this computational turn mean for traditional modes of content analysis? Specifically, I consider the traditional (manual) approach of conducting a content analysis–a primary method in the study of media messages–in light of the proliferation of computer-centric approaches, assess what is gained and lost in turning to predominantly computational solutions, and discuss an alternative approach that aims to effectively combine traditional and computational modes to facilitate more expansive and powerful–yet still reliable and meaningful–analyses of media content.

Bio:  Rodrigo Zamith is a doctoral candidate in the School of Journalism and Mass Communication at the University of Minnesota. His research focuses on the reconfiguration of journalism and the development of digital research methods. His work has been published in the Journal of Computer-Mediated Communication, the Journal of Broadcasting and Electronic Media, and Digital Journalism.

David Jensen

David Jensen

School of Computer Science, University of Massachusetts
Using Graphical Models to Reason About Quasi-Experimental Designs
Friday, October 10, 2014 • 12:00PM-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Effective methods for inferring causal dependence from observational data have been developed within both computer science and quantitative social science. Methods in computer science have focused on the correspondence between casual graphical models and observed patterns of statistical association. Methods in social science have focused on templates for causal inference often called quasi-experimental designs, including designs that use instrumental variables, propensity scores matching, regression discontinuity, and interrupted time-series. In this talk, I will describe many of the known experimental and quasi-experimental designs in the language of directed graphical models, and I will show how the graphical model framework allows effective reasoning about threats to validity in these designs. Finally, I will present two novel designs that have resulted from our recent work on causal inference in relational data.

Bio:  David Jensen is Associate Professor of Computer Science and Director of the Knowledge Discovery Laboratory at the University of Massachusetts Amherst. He received his doctorate from Washington University in St. Louis in 1992. From 1991 to 1995, he served as an analyst with the Office of Technology Assessment, an agency of the United States Congress. His research focuses on machine learning and causal inference in relational data sets, with applications to social network analysis, computational social science, fraud detection, and management of large technical systems. He has served on the Executive Committee of the ACM Special Interest Group on Knowledge Discovery and Data Mining and on the program committees of the International Conference on Machine Learning, the International Conference on Knowledge Discovery and Data Mining, and the Uncertainty in AI Conference. He was a member of the 2006-2007 Defense Science Study Group, and served for six years on DARPA's Information Science and Technology (ISAT) Group. He is the incoming Associate Director of the UMass Computational Social Science Institute. He won the 2011 Outstanding Teaching Award from the UMass College of Natural Science.

Krista Gile

Krista Gile

Assistant Professor, Department of Mathematics and Statistics
Inference and Diagnostics for Respondent-Driven Sampling Data
Friday, October 17, 2014 • 12:00pm-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Respondent-Driven Sampling is type of link-tracing network sampling used to study hard-to-reach populations. Beginning with a convenience sample, each person sampled is given 2-3 uniquely identified coupons to distribute to other members of the target population, making them eligible for enrolment in the study. This is effective at collecting large diverse samples from many populations.
Unfortunately, sampling is affected by many features of the network and sampling process. In this talk, we present advances in sample diagnostics for these features, as well as advances in inference adjusting for such features.

Bio:  Krista J. Gile is Assistant Professor of Statistics at UMass Amherst. Her research focuses on developing statistical methodology for social and behavioral science research, particularly related to making inference from partially-observed social network structures. Most of her current work is focused on understanding the strengths and limitations of data sampled with link-tracing designs such as snowball sampling, contact tracing, and respondent-driven sampling.

Jesse Rhodes

Jesse Rhodes

Associate Professor, Department of Political Science
The Politics of Class-Based Appeals in American Presidential Campaigns
Friday, October 24 2014 • 12:00pm-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  In politics, discussion of class is inexorably linked to matters relating to the distribution and redistribution of wealth. Consequently, class appeals are a powerful - but politically perilous - form of campaign rhetoric. Because class terms are heavily freighted with meaning and most Americans identify with one economic class or another, explicit appeals to class identities may serve to mobilize targeted groups in elections. However, because class groups that are mis-targeted may punish the speaker, candidates must take care in crafting class appeals.
Drawing on a new dataset of every explicit class reference by Democratic and Republican presidential candidates between 1952 and 2012, I use a variety of quantitative text analytic methods to examine the volume and topical content of candidates' class appeals, and account for variation in content between candidates and over time. The study sheds new light on the factors affecting presidential candidates' decisions to attempt to mobilize voters on the basis of class.

Bio:  Jesse Rhodes is associate professor in the Department of Political Science at the University of Massachusetts, Amherst. He maintains research and teaching interests in the areas of the American presidency, party politics, social policy, and American political development. His methods encompass a wide range of quantitative and qualitative methods, including large-scale text collection and analysis, survey research, interviewing, archival research, and historical analysis.

Justin H. Gross

Justin H. Gross

Assistant Professor of Political Science, UNC-Chapel Hill
Building Idea-oriented Measures of Ideology in Text
Tuesday, November 25 2014 • 10:00AM-11:30AM
Campus Center Room 805-809

Abstract:  In political communication, philosophy, and psychology, as well as in everyday life, the word "ideology" carries various different meanings. When working with text written by political leaders, members of opinion media, and others writing explicitly about their views and beliefs, we have a unique opportunity to study ideology directly as publicly articulated political philosophy (however crudely expressed). Using a corpus of over two hundred popular books by contemporary American ideologues, my coauthors and I develop two approaches to measurement that are oriented toward detecting differences in recognizable viewpoints. One is a two stage approach that first identifies a set of terms that are useful in distinguishing among recognized ideological classes, then represents new documents via hidden Markov model in order to estimate how often a speaker sounds like someone in each prototypical class. In the second approach, we begin by carefully defining key concepts and then use the corpus to identify words, phrases, and Boolean rules that are sensitive markers of the concepts, allowing us to measure the relative attention writers devote to abstract ideas on which ideologies are built, what we refer to as their "ideational agendas." I discuss and illustrate each method and identify challenges to effective implementation.

Bio:  Justin H. Gross is Assistant Professor of Political Science at UNC-Chapel Hill. His research interests include problems of measurement in networks, text, and survey data. Substantively, he is interested in political communication, particularly the competitive framing of policy issues by opinion media. He is currently working with computer scientists and social scientists on the development of semi-supervised techniques of frame detection for the study of large databases of news stories.

Mark Pachucki

Mark Pachucki

Affiliated Faculty, Division of General Pediatrics,Massachusetts General Hospital
Instructor of Medicine and Pediatrics, Harvard Medical School
Physical activity and social influence in early adolescent networks: How measurement matters
Tuesday, December 2, 2014 • 11:30AM-1:00PM
Machmer W32

Abstract:  Physical activity (PA) is a modifiable health behavior that has been associated with cardiometabolic disorder, cancer, and distressed mental health. Among older youth it has been consistently shown that friends tend to be similar in their PA levels. However, research examining why this is the case has revealed inconsistencies in theorized mechanisms responsible for this similarity. Youth tend to form friendships based upon existing PA behaviors, but one’s peer group can also influence changes in a youth’s own PA. Given this context, surprisingly little research has explored how social relationships may shape PA during early adolescence considered as a discrete stage of the life course. Relative to childhood or late adolescence, early adolescence has unique properties in terms of social, psychological, biological, and neurological development that interact to shape decisions around health behaviors. Moreover, the majority of PA network research relies on PA and social relationships data obtained from self-report. In this paper, we compare early adolescent peer effects on accelerometer-measured PA using both cognitive and high-quality behavioral affiliation measures. Data were obtained by recruiting an unusually complete cohort of 6th-grade students and observing their behaviors at multiple points during a four-month period. Findings suggest that replacing network self-report with behavioral measures of social interaction yields a great deal more precision with lower respondent burden, but also that each type of network measurement has unique insights to offer.

Bio:  Mark C. Pachucki investigates how interpersonal relationships can shape health behaviors and outcomes across the life course. He is jointly appointed as Senior Scientist at the Mongan Institute for Health Policy at Massachusetts General Hospital and Affiliated Faculty with the Massachusetts General Hospital for Children Division of General Pediatrics, with academic appointments as Instructor in Medicine and Pediatrics at Harvard Medical School. His training is in sociology and social determinants of health, and current projects focus on relationship formation and dissolution and how peer and family relationships - specifically, between children and their peers, between parents and children, and between spouses - can influence the adoption of risky health behaviors. Pachucki has recently published his research in Social Science & Medicine, Annual Review of Sociology, Poetics, Sociologie et sociétés, the International Journal of Obesity, American Journal of Preventive Medicine, and American Journal of Public Health.

Alex Hanna

Alex Hanna

Ph.D. Candidate, Department of Sociology, University of Wisconsin
Developing a System for the Automated Coding of Protest Event Data
Friday, December 5, 2014 • 12:00pm-2:00PM
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Scholars and policy makers recognize the need for better and timelier data about contentious collective action, both the peaceful protests that are understood as part of democracy and the violent events that are threats to it. News media provide the only consistent source of information available outside government intelligence agencies and are thus the focus of all scholarly efforts to improve collective action data. Human coding of news sources is time-consuming and thus can never be timely and is necessarily limited to a small number of sources, a small time interval, or a limited set of protest "issues" as captured by particular keywords. There have been a number of attempts to address this need through machine coding of electronic versions of news media, but approaches so far remain less than optimal. The goal of this paper is to outline the steps needed to build, test and validate an open-source system for coding protest events from any electronically available news source using advances from natural language processing and machine learning. Such a system should have the effect of increasing the speed and reducing the labor costs associated with identifying and coding collective actions in news sources, thus increasing the timeliness of protest data and reducing biases due to excessive reliance on too few news sources. The system will also be open, available for replication, and extendable by future social movement researchers, and social and computational scientists.

Bio:  Alex Hanna is a PhD candidate in sociology at the University of Wisconsin-Madison. Substantively, they are interested in social movements, media, and the Middle East. Methodologically, they are interested in computational social science, textual analysis, and social network analysis. Alex's work has appeared in both social and computational science venues, including Mobilization, the ANNALS of the American Academy of Political and Social Science, and ICWSM. They also co-founded and contribute regularly to the computational social science blog Bad Hessian, where they write about Python, R, and Twitter.