Computational Social Science Institute

University of Massachusetts Amherst

Lecture series generously sponsored by Yahoo!

Videos of some CSSI seminars are available  here

Upcoming

Friday, September 30, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151

Faculty and graduate students come to a research panel and reception to celebrate and welcome 15 new CSSI affiliates (including 9 new hires) who joined CSSI in the past year. Today, 7 new affiliates will give brief introductions to their research interests, and we will spend the rest of the time sharing lunch, mixing and mingling across colleges.

The following affiliates will present:
Joshua Brown, Journalism
Song Gao, Civil & Environmental Engineering
Jenna Marquard, Mechanical & Industrial Engineering
Shannon Roberts, Mechanical and Industrial Engineering
Rong Rong, Resource Economics
Doug Rice, Political Science
Florence Sullivan, Education

Tina Eliassi-Rad

Tina Eliassi-Rad

Department of Computer Science, Rutgers University
Theoretical and Empirical Guides for Selecting Algorithms on Complex Networks 

Friday, October 7, 2016 • 12:30 p.m.-2:00 p.m.
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 discuss two problems on complex networks. (1) Measuring tie-strength: Given a set of people and a set of events attended by them, how should we measure connectedness or tie strength between each pair of persons? The underlying assumption is that attendance at mutual events produces an implicit social network between people. I will describe an axiomatic solution to this problem. (2) Network similarity: Given two networks (without known node-correspondences), how should we measure similarity between them? This problem occurs frequently in many real-world applications such as transfer learning, re-identification, and change detection. I will present an empirical guide on how to select a network-similarity method.

Bio:  Tina Eliassi-Rad is an Associate Professor of Computer Science at Rutgers University. Before joining academia, she was a Member of Technical Staff and Principal Investigator at Lawrence Livermore National Laboratory. Tina earned her Ph.D. in Computer Sciences (with a minor in Mathematical Statistics) at the University of Wisconsin-Madison. Her current research lays at the intersection of graph mining, network science, and computational social science. Within data mining and machine learning, Tina's research has been applied to the World-Wide Web, text corpora, large-scale scientific simulation data, complex networks, fraud detection, and cyber situational awareness. She has published over 60 peer-reviewed papers (including a best paper runner-up award at ICDM'09 and a best interdisciplinary paper award at CIKM'12); and has given over 120 invited presentations. Her algorithms have been incorporated into systems used by the government and industry (e.g., IBM System G Graph Analytics) as well as open-source software (e.g., Stanford Network Analysis Project). In 2010, she received an Outstanding Mentor Award from the US DOE Office of Science. For more details, visit http://eliassi.org.

Dunia López-Pintado

Dunia López-Pintado

Economics Department, Universidad Pablo de Olavide, Seville, Spain
Diffusion in Social and Economic Complex Networks 

Friday, October 14, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Some behaviors, ideas or technologies spread and become persistent in society, whereas others vanish. This paper analyzes the role of social influence in determining such distinct collective outcomes. We characterize, as a function of the primitives of the dynamic model, the diffusion threshold (i.e., the spreading rate above which the adoption of the new behavior becomes persistent in the population) and the endemic state (i.e., the fraction of adopters in the stationary state of the dynamics). The results highlight the importance of the correlation between visibility and information for diffusion purposes. Moreover, the level of spread of a (positive or negative) behavior in a population can be largely affected by whether interactions occur mainly between individuals of the same type (i.e., there is homophily or assortativity) or between individuals of different types (i.e., there is group mixing or dissasortativity). We find that there can be levels of mixing or segregation that are inefficient, i.e., undesirable for all types. We also discuss the existence of optimal levels of segregation focusing on average diffusion.

Bio:  Dunia López-Pintado is Assistant Professor of Economics at Universidad Pablo de Olavide (Seville, Spain) with an affiliation to the Center for Operations Research and Econometrics (Universite Catholique de Louvain, Belgium). She received her PhD in Economics in the Universidad de Alicante (Spain). She worked as a post-doc in the Social and Information Science Laboratory in Caltech and the Collective Dynamics Group in Columbia University. She applies tools from statistical physics and game theory to study various topics in economics and sociology including diffusion in network, group incentives and local public goods. She has published in journals such as Games and Economic Behavior, Rationality and Society, and Network Science.

Weiai Wayne Xu

Weiai Wayne Xu

Department of Communication, Unoiversity of Massachusetts Amherst
www.curiositybits.com
Bridging the Data Gap - Challenges in Applying Computational Approaches to Communication Research 
Co-sponsored by the Department of Communication.

Friday, October 21, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  As more and more communication researchers are embracing data science, they oftentimes encounter challenges in getting around restrictive APIs, filtering noises through unstructured digital behavioral data, and most importantly, connecting analytics to theories and contexts. In this talk, I will present my conversations with communication scholars and share their concerns about the accessibility and validity of computational approaches in communication research, in particular, qualitative research. I will also speak from my own research experience, and propose that data scientists develop a social media analytics toolkit accessible to communication researchers and designed with the purpose of guiding qualitative analyses in later stages of research.

Bio:  Weiai Xu (Wayne) is an Assistant Professor in the Department of Communication at University of Massachusetts Amherst. Wayne previously worked at Network Science Institute at Northeastern University as a postdoc researcher. He earned his Ph.D. in Communication from SUNY-Buffalo. Wayne 's research examines online communities and their implications for political engagement and message diffusion. His research has appeared in various journals, including American Behavioral Scientist, Journal of Broadcasting & Electronic Media, Online Information Review, Quality & Quantity, International Journal of Communication. He has also assisted three national grant projects in the areas of online public opinion and online community engagement. Wayne's research at UMass seeks to identify communication strategies to connect divided online groups.

Kelly Joyce

Kelly Joyce

Director, Science, Technology, and Society program, Drexel University
Algorithms, Big Data, Disciplinary Expertise and Inequalities 
Co-sponsored with the Department of Sociology and the Institute for Social Science Research
Note non-standard date, time, and location
Please RSVP for this event through the Institute for Social Science Research

Thursday, October 27, 2016 • 4:00 p.m.
107 Bartlett Hall

Abstract:  In this seminar, Kelly Joyce (Professor of Sociology and Director of the Science, Technology, and Society program | Drexel University) will discuss how disciplinary expertise shapes the values that drive big data work and the implications of this process for understanding Inequalities. The talk draws on NSF-funded fieldwork and interviews conducted at four field sites where research teams aim to develop health-related big data. It shows the importance of social science expertise to big data work.

Bio:  TBA

Title: TBA
Abstract: TBA
Friday, November 4, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
November 18, 2016
November 25, 2016
Title: TBA
Abstract: TBA
Friday, December 2, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Title: TBA
Abstract: TBA
Friday, December 9, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151

Fall 2016

Balázs Kovács

Balázs Kovács

Organizational Behavior, Yale School of Management
Selective Rating Behavior Undermines the Wisdom-of-the-Crowd for Average Ratings 

Friday, September 23, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  A key purpose of the average ratings displayed on recommendation websites is to guide decision-making. The wisdom-of-the-crowd mechanism suggests average ratings provide unbiased quality estimates. Yet, we show that average ratings are negatively biased, especially those based on few ratings. This is due to an asymmetry in the dynamics of average ratings: higher average ratings change more frequently than lower average ratings because high average ratings attract more additional ratings. Crucially, this asymmetry implies a spurious association between number of ratings and average ratings. We find evidence for these patterns of evaluative biases in analyses of rating data from Amazon and Yelp and in a randomized experiment. Even if individual ratings are faithful representations of the experiences of users, average ratings will be systematically biased.

Bio:  Balazs Kovacs is an Assistant Professor of Organizations and Management at the Yale School of Management, with faculty affiliations to the Yale Networks Science Institute and the Yale Department of Sociology. Prior to coming to Yale, he was an Assistant Professor at the University of Lugano, Switzerland. He received his PhD in Organizational Behavior and MA in Sociology from Stanford University. He studies various topics in organization theory and sociology, including social networks, learning, diffusion, innovation, identity, culture, and status. He has published in leading management and sociology journals such as Administrative Science Quarterly, American Sociological Review, Management Science, Organization Science, and Social Networks. His research has been covered in major media outlets such as the Guardian and the New York Times.

Kim Geissler

Kim Geissler

Department of Health Promotion and Policy, University of Massachusetts
Coordination Within Teams and The Cost of Health Care 
(joint paper with Leila Agha, Keith Ericson, Benjamin Lubin, and James Rebitzer)

Friday, September 16, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  We examine how primary care physicians (PCPs) assemble teams of specialists to care for their patients. In our model, PCPs can invest in a relationship with specialists that requires upfront costs but has benefits for care coordination. PCPs who work with fewer specialists (have higher "referral concentration") invest more in relationship-specific capital. Using the Massachusetts All Payer Claims Database of health insurance enrollment and claims, we show that this team-based coordination of care measure is virtually uncorrelated with existing patient-based coordination of care measures. We identify the effect of referral concentration on spending by comparing the spending of individuals who see the same specialist, but come from PCPs with different referral concentration. We use the same technique with standardized prices to distinguish whether the effects are a result of specialist prices or utilization effects. We find that a one standard deviation increase in coordination of care by a PCP reduces average costs by 2.2%.

Bio:  I am a health policy researcher using health insurance claims datasets and advanced empirical methods to examine factors affecting access to and quality of health care. My current work focuses on physician referrals and how patient care is coordinated among different physicians treating the same patient. To empirically measure these referrals, I use health insurance claims from nearly the entire non-elderly insured population in Massachusetts. I use economic modeling and social network analysis to examine the effects of organizational affiliations--and changes in these affiliations due to hospital mergers and acquisitions–-on physician referral patterns. I also analyze the impacts of such patterns on health care cost and quality.

Chris Danforth

Chris Danforth

Department of Mathematics & Statistics, University of Vermont
Limits to Socio-Cultural Inference from Tweets & Books 

Friday, September 9, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Scientific analysis of large-scale text has begun to reveal remarkable insights into human behavior. Indeed, there is growing evidence that our society’s daily online interactions can be appropriately aggregated into digital measures of physical mobility, emotional health, and linguistic evolution. However, the socio-technical instruments of tomorrow will be limited by the quality of the data they are fed, as well as human awareness of their existence. This talk will describe our ongoing effort to quantify the population-scale sentiment associated with any topic using Twitter, including a public health application where we found 80% of opinions to be expressed by non-human actors. We will also describe our recent analysis of the Google Books corpus, and point to the need to fully characterize its behavior before drawing broad conclusions about cultural dynamics. Finally, we will describe a proof-of-concept study using photographic data from Instagram to predictively screen for depression.

Bio:  Chris Danforth is the Flint Professor of Mathematical, Natural, and Technical Sciences at the University of Vermont. With colleague Peter Dodds, he co-directs the Computational Story Lab, a group of applied mathematicians and data scientists at the undergraduate, masters, PhD, and postdoctoral level working on large-scale, systems problems in many fields including sociology, nonlinear dynamics, networks, ecology, and physics. The group has built several socio-technical instruments including the Hedonometer and the Lexicocalorimeter. Danforth's formal background is in nonlinear dynamics applied to weather and climate prediction, and he is a member of the Mathematics & Climate Research Network.

Spring 2016

Joe DiGrazia

Joe DiGrazia

Neukom Fellow, Neukom Institute, Dartmouth College
Internet Search Data as a Measure of Behaviors and Attitudes in Social Science Research 

Friday, April 29, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  This paper examines the utility of using aggregate Internet search data from online search engines, such as Google, to construct state-level measures of attitudes and behaviors in the United States. Generally, state-level research relies on demographic statistics, official statistics produced by government agencies, or aggregated survey data. However, each of these data sources has serious limitations in terms of both the availability of the data and its ability to capture important concepts. Using several different case studies, this paper demonstrates the ability of aggregate search data to capture important dimensions of attitudes and behaviors that are notoriously difficult to measure using traditional data sources. First, state-level search measures for anti-immigrant sentiment and economic distress are developed and compared to traditional metrics that are typically used to measure these concepts, like the unemployment rate and the international immigration rate in their ability to successfully predict Tea Party event counts between 2009 and 2011. The results show that the Google search measures are effective at predicting Tea Party mobilization in a way that is consistent with existing theory, while the traditional measures are not. Second, state-level search measures are employed to measure interest in two popular sets of political conspiracy theories: those relating to the “Illuminati” and those relating to the citizenship of President Obama across US states between the years 2007 and 2014. Drawing on previous literature which has found that individuals tend to engage in conspiratorial ideation when experiencing feelings of threat or insecurity, the analysis looks at potential social sources of insecurity or threat in the social and political environment that might motivate conspiratorial ideation, including partisan changes in control of the presidency, immigration, and unemployment.

Bio:  I am a postdoctoral fellow in the Department of Sociology and the Neukom Institute for Computational Science at Dartmouth College. My work focuses on political sociology and social movements, as well as developing novel data sources from the digital traces created by the public’s use of communication and information technology. My work has been published in sociology and communications journals such as Sociological Methods & Research and Information, Communication & Society and has been covered in media outlets such as TIME Magazine and the Columbia Journalism Review.

April 22, 2016 • 8:30 a.m.-5:00 p.m.
Computer Science Building

The UMass Amherst Center for Data Science will be hosting its first Annual Data Science Research Symposium. Join us to interact with our faculty and senior researchers as they present their most recent research and industry participants share accomplishments and challenges. Discuss technical trends and projections in breakout sessions on focused topics, such as machine learning & optimization, scalable data management, human language technology, computer vision, digital healthcare & biomedicine and business analytics, finance & insurance. Engage data science graduate students during a poster session highlighting new research.

For additional details and to register please visit ds.cs.umass.edu/event/data-science-research-symposium.

Co-sponosored with the Institute for Social Science Research
Friday, April 8, 2016 • 12:30 - 2:00 p.m.
A light lunch will be served at 12:30
Bartlett 107

Abstract: The ability to replicate research findings is an essential component of the scientific process. However, the scientific process itself has come under considerable scrutiny due to recent evidence that the results of many studies and experiments are difficult, if not impossible, to replicate. This panel discussion will explore the causes, controversies and consequences of the "Replication Debate" featuring methodologists and researchers from across the social and computational sciences.

Panelists (left-to-right): Emery Berger (Information and Computer Sciences), Thomas Herndon (Economics), David Jensen (Information and Computer Sciences), Caren Rotello (Psychological and Brain Sciences), Adrian Staub (Psychological and Brain Sciences)

Julia Lane

Julia Lane

New York University, Center for Urban Science Progress (CUSP)
What Social Science Research Brings to Big Data Research: A View From Experience 
Colloquium co-sponsored with ISSR and the Department of Economics.

Friday, April 1, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  The world of big data offers enormous opportunities for social science, but social scientists have a great deal to offer, to a (data) world that is currently looking to computer scientists to provide answers. Three major areas in which social scientists can contribute, based on decades of experience and work with end users, include causal inference, data quality and addressing privacy and confidentiality. This talk discusses each of these in turn using examples drawn from a new largescale social science data infrastructure constructed using big data techniques.

Bio:  Julia Lane is a Professor in the Wagner School of Public Policy at New York University. She is also a Provostial Fellow in Innovation Analytics and a Professor in the Center for Urban Science and Policy at NYU. Dr. Lane is an economist and has authored over 65 refereed articles and edited or authored seven books. She has been working with a number of national governments to document the results of their science investments. Her work has been featured in several publications including Science and Nature. Dr. Lane started at the National Science Foundation (as Senior Program Director of the Science of Science and Innovation Policy Program) to quantify the results of federal stimulus spending, which is the basis of the new Institute for Research on Innovation and Science at the University of Michigan. Dr. Lane has had leadership positions in a number of policy and data science initiatives at her other previous appointments, which include Senior Managing Economist at the American Institutes for Research; Senior Vice President and Director, Economics Department at NORC/University of Chicago; various consultancy roles at The World Bank; and Assistant, Associate and Full Professor at American University. Dr. Lane received her PhD in Economics and Master’s in Statistics from the University of Missouri.

Friday, March 25 • 12:00PM–3PM
Campus Center -- Invite-only Please
[There is no regular CSSI seminar this week.]
Chris Bail

Chris Bail

Department of Sociology, Duke University
http://www.chrisbail.net/
Cultural Flow: How Autism Advocacy Organizations Stimulate Public Conversation on Facebook 
Note non-standard date, time, and location

Thursday, March 10, 2016 • 11:30 p.m.-1:00 p.m.; lunch will be served
Thompson Hall room 919

Abstract:  This paper presents a new method for automated text analysis that draws upon social network analysis in order to establish what types of discourses produced by Autism Advocacy Organizations stimulate public comments/conversation on Facebook. I offer an ecological theory that suggests organizations succeed based upon the relationship between their discourses and the broader field of discourse about a topic such as Autism Spectrum Disorders. I also present an app-based research design that enables collection of both public and private Facebook data alongside conventional survey research methods.

Bio:  Chris Bail is an Assistant Professor of Sociology at Duke University. He studies how non-profit organizations and other political actors shape public discourse by analyzing large groups of texts from newspapers, television, public opinion surveys, and social media sites such as Facebook and Twitter. His research has been published by Princeton University Press, American Sociological Review, Sociological Theory, Theory and Society, and Sociological Methods and Research. His work has been recognized by awards from the American Sociological Association, the Association for Research on Non-Profit Organizations and Voluntary Action, the Society for the Scientific Study of Religion, and the Society for Study of Social Problems, and supported by the National Science Foundation and the Robert Wood Johnson Foundation. His research has also been covered by major media outlets such as NBC News, National Public Radio, and the Washington Post. Bail earned his Ph.D. from Harvard University in 2011.

March 4, 2016
Computer Science graduate student recruitment event
Benjamin Marlin

Benjamin Marlin

Assistant Professor in the School of Computer Science, University of Massachusetts Amherst
The PERSPeCT Project: Recommender Systems for Personalizing Health Behavior Change Interventions 

Friday, February 26, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Web-based recommender systems match online content like web pages, news articles, videos, and music to the preferences and interests of individual users. The aim of the PERSPeCT project is to incorporate these same methods into a computer tailored health communications (CTHC) intervention system with the goal of more effectively eliciting positive health behavior changes through improved personalization. Most current CTHC systems operate by selecting and sending motivational messages to patients according to a fixed set of expert-derived rules. By contrast, PERSPeCT collects feedback about each message sent to a patient and uses this feedback to personalize the selection of future messages using probabilistic inference in a Bayesian latent variable model. In this talk, I will present the preliminary studies that informed the development of the PERSPeCT system, including a study of feedback rating semantics in the CTHC domain, and the empirical evaluation of a number of classical and recent models applied to CTHC message ratings. I will conclude by describing the results of a clinical trial of the PERSPeCT system. This is joint work with Dr. Thomas Houston's research group in the Department of Quantitative Health Sciences at the UMass Medical School.

Bio:  Benjamin Marlin joined the University of Massachusetts Amherst as an assistant professor of Computer Science in fall 2011 where he co-founded and co-directs the Laboratory for Machine Learning and Data Science. His current research interests focus on machine learning models and algorithms for time series data analysis with an application focus in health and behavioral science. Marlin is a member of the UMass Amherst Computational Social Science Institute, the Center for Intelligent Information Retrieval, and the Institute for Applied Life Sciences. He is also a member of the leadership team for the the Center for Data Science. Marlin received an NSF CAREER award in 2014, and a Yahoo! Faculty Research Engagement Program award in 2013. Prior to joining UMass Amherst, Marlin was a fellow of the Pacific Institute for the Mathematical Sciences and the Killam Trusts at the University of British Columbia. He received his PhD in machine learning from the University of Toronto.

Data Science Talk: Robert West

Data Science Talk: Robert West

Stanford University Infolab
Human Behavior in Networks 
Colloquium co-sponsored with the Center for Data Science and College of Information and Computer Science
Please contact host Brendan O'Connor (brenocon@cs.umass.edu) if interested in scheduling a meeting

Thursday, February 18, 2016 • 1:00 p.m.-2:00 p.m.
Computer Science Building, Room 150/151

Abstract:  Humans as well as information are organized in networks. Interacting with these networks is part of our daily lives: we talk to friends in our social network; we find information by navigating the Web; and we form opinions by listening to others and to the media. Thus, understanding, predicting, and enhancing human behavior in networks poses important research problems for computer and data science with practical applications of high impact. Navigation constitutes one fundamental human behavior: in order to make use of the information and resources around us, we constantly explore, disentangle, and navigate networks such as the Web. Studying navigation patterns lets us understand better how humans reason about complex networks and lets us build more human-friendly information systems. As an example, I will present an algorithm for improving website hyperlink structure by mining raw web server logs. The resulting system is being deployed on Wikipedia's full server logs at terabyte scale, producing links that are clicked 10 times as frequently as the average link added by Wikipedia editors. Communication and coordination through natural language is another prominent human network behavior. Studying the interplay of network structure and language has the potential to benefit both sociolinguistics and natural language processing. Intriguing opportunities and challenges have arisen recently with the advent of online social media, which produce large amounts of both network and natural-language data. As an example, I will discuss my work on person-to-person sentiment analysis in networks, which combines the sociological theory of structural balance with techniques from natural language processing, resulting in a sentiment prediction model that clearly outperforms both text-only and network-only versions. I will conclude the talk by sketching interesting future directions for computational approaches to studying human behavior in networks.

Bio:  Robert West is a sixth-year Ph.D. candidate in Computer Science in the Infolab at Stanford University, advised by Jure Leskovec. His research aims to understand, predict, and enhance human behavior in social and information networks by developing techniques in data science, data mining, network analysis, machine learning, and natural language processing. Previously, he obtained a Master's degree from McGill University in 2010 and a Diplom degree from Technische Universitat Munchen in 2007.

Elizabeth Ogburn

Elizabeth Ogburn

Johns Hopkins Bloomberg School of Public Health, Department of Biostatistics
Inference in the Presence of Network Dependence Due to Contagion 

Friday, February 12, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Interest in and availability of social network data has led to increasing attempts to make causal and statistical inferences using data collected from subjects linked by social network ties. But inference about all kinds of estimands, starting with simple sample means, is challenging when only a single network of non-independent observations is available. There is a dearth of principled methods for dealing with the dependence that such observations can manifest. We describe methods for causal and semiparametric inference when the dependence is due solely to the transmission of information or outcomes along network ties.

Bio:  My research is in causal inference and epidemiologic methods. Broadly, I am interested in developing methods for and describing the behavior of traditional statistical machinery when standard assumptions are not met. I have worked on characterizing the bias that results from misclassification, i.e., violations of the assumption that variables were measured accurately. I have also worked on semiparametric estimation of instrumental variables models, as these models are useful for certain violations of “no unmeasured confounding” assumptions. Currently, my main focus is on developing new methods for statistical and causal inference in the presence of interference (when one subject’s treatment may affect other subjects’ outcomes) and for social network data; both of these represent violations of assumptions of independence among observations.

Justin Gross

Justin Gross

Department of Political Science
Connected by Comments: The Evolving Affect Network of Presidential Candidates on Twitter During the 2016 GOP Nomination Contest 
Cosponsored by Department of Political Science

Friday, January 29, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  The unprecedented number of serious candidates in the Republican Party’s 2016 nomination contest for the U.S. Presidency provides a rare opportunity to examine the changing nature of affective relationships among candidates. All seventeen major candidates have Twitter accounts and have tweeted comments about their opponents both before the campaigning began and over the course of the campaign for the GOP nomination. Political scientists, writing on the phenomenon of negative campaigning, have made a number predictions about what conditions will make it more likely that a campaign shall “go negative.” These researchers have concentrated on advertisements, but the number and variety of advertisements produced are highly dependent on a campaign’s resources. By contrast, candidates’ use of social media allows us to directly observe, in real time, candidate interaction. Furthermore, the nature of “going negative” is more complicated in a crowded field and would seem to benefit from a network analytic approach. Structural balance theory, in particular, provides some guidance in thinking about the dynamics of positive and negative affect, operationalized as a signed network. However, peer group and organizational behavior, commonly driving theories of structural balance, are rather distinct from the behavior among electoral competitors. How likely is it that, in an environment tending toward mutual animosity, a taste for structural balance will somehow prevail? I address this question, examine clustering patterns, and describe some highlights of the online conflicts and collaborations that have played out this election season.

Bio:  Justin H. Gross is Assistant Professor of Political Science at UMass Amherst. He holds a Ph.D. in Statistics and Public Policy from Carnegie Mellon University. His applied research interests are in mass media and political communication, public opinion, and public policy. He works on methodological problems in measurement, text analysis, and network analysis, and is especially interested in methods that put statistical and computational tools to use in service of our ability to achieve rich qualitative insights. Recently, he has been collaborating with a cross-disciplinary team of political scientists and computational linguists, developing tools for the detection of issue frames and publicly expressed ideologies in text.

David Huber

David Huber

Department of Psychological & Brain Sciences
The Rise and Fall of the Recent Past: A Unified Account of Immediate Repetition Paradigms 

Friday, January 22, 2016 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Perception does not happen instantaneously. Instead, perceptual information is accumulated gradually (i.e., a rise in the perceptual response) to achieve accurate identification despite perceptual noise. This temporal integration of information predicts that previous presentations may become erroneously blended with subsequent presentations. An optimal decision process can reduce this source confusion by discounting perceptual evidence that may have come from previous presentations. Furthermore, habituation (i.e., a fall in the perceptual response) may be the brain’s trick for approximating this optimal decision. Habituation reduces blending and enhances change detection: because previous perceptions are habituated, anything new stands out. However, this solution comes with a cost, making it difficult to detect immediate repetitions. Over the last decade, I have tested this theory, examining its application to a range of different behavioral tasks that involve immediate repetitions. In this talk I present a new line of experiments addressing the consequences of this theory for dynamic shifts of attention and the finding that it is difficult to allocate attention more than once within a short time period. In brief, "inhibition of return" is explained as a repetition deficit for detecting where something occurs and the "attentional blink" is explained as a repetition deficit for detecting when something occurs.

Bio:  David Huber is a professor in the Department of Psychological and Brain Sciences at the University of Massachusetts, Amherst. He is the director of the Cognitive Experiments, Models, and Neuroscience Laboratory (CEMNL), which focuses on human perception and memory from a broad-based, computational perspective. To shed light on these basic cognitive processes, his research finds converging evidence from behavioral studies and neurophysiological measures in combination with neural network and Bayesian modeling. Ongoing research topics include recognition/recall memory, the benefits of retrieval practice, metamemory, letter/word perception, face perception, semantics, shifts of attention, and social cognition.

CSSI talks hosted by the Department of Resource Economics:

Ana Gazmuri, Wharton School
School Segregation in the Presence of Student Sorting and Cream-Skimming: Evidence from a School Voucher Reform
Thursday, January 14, 2016 • 10:00 a.m. • Stockbridge Hall Room 303
Jiaxuan Li, Boston University
Gateway Products in the DSLR Camera Market: Dynamic Demand, Consumer Learning and Switching Costs
Friday, January 15, 2016 • 10:00 a.m. • Stockbridge Hall Room 303
Fedor Iskahov, University of New South Wales
Recursive Lexicographical Search: Finding All Markov Perfect Equilibria of Finite State Directional Dynamic Games
Tuesday, January 26, 2016 • 9:45 a.m. • Stockbridge Hall Room 303
Brian Adams, California State University, East Bay
Zone Pricing and Strategic Interaction: Evidence from Drywall
Wednesday, January 27, 2016 • 11:00 a.m. • Stockbridge Hall Room 303
Debi Mohapatra, Cornell University
Price Control and Access to Drugs: The Case of India's Malarial Market
Friday, January 29, 2016 • 10:00 a.m. • Stockbridge Hall Room 303

Fall 2015

CSSI talks hosted by the Department of Communication:

Weiai (Wayne) Xu
Big Data, Small Data and Everything in Between
Monday, November 30, 2015 • 9:00 a.m.–10:15 a.m. • Campus Center, Room 903
Marco Bastos
Cross-effects Between Online and Offline Social Networks
Monday, December 7, 2015 • 9:00 a.m.–10:15 a.m. • Campus Center, Room 911-915
Syed Saif Shahin
Title: Van Rossum meets Van Dijk: Using Computational Tools for Critical Research
Friday, December 11, 2015 • 9:00 a.m.–10:15 a.m. • Campus Center, Room 903
David Blei

David Blei

Columbia University
Probabilistic Topic Models and User Behavior 
Cosponsored by Center for Data Science and Mathematics & Statistics
Note different location

Friday, November 20, 2015 • 12:30-p.m.-2:00 p.m.
Lederle Graduate Research Tower Room 1634
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Topic modeling algorithms analyze a document collection to estimate its latent thematic structure. However, many collections contain an additional type of data: how people use the documents. For example, readers click on articles in a newspaper website, scientists place articles in their personal libraries, and lawmakers vote on a collection of bills. Behavior data is essential both for making predictions about users (such as for a recommendation system) and for understanding how a collection and its users are organized.

I will review the basics of topic modeling and describe our recent research on collaborative topic models, models that simultaneously analyze a collection of texts and its corresponding user behavior. We studied collaborative topic models on 80,000 scientists' libraries from Mendeley and 100,000 users' click data from the arXiv. Collaborative topic models enable interpretable recommendation systems, capturing scientists' preferences and pointing them to articles of interest. Further, these models can organize the articles according to the discovered patterns of readership. For example, we can identify articles that are important within a field and articles that transcend disciplinary boundaries.

Bio:  David Blei is a Professor of Statistics and Computer Science at Columbia University, and a member of the Columbia Data Science Institute. His research is in statistical machine learning, involving probabilistic topic models, Bayesian nonparametric methods, and approximate posterior inference algorithms for massive data. He works on a variety of applications, including text, images, music, social networks, user behavior, and scientific data. David has received several awards for his research, including a Sloan Fellowship (2010), Office of Naval Research Young Investigator Award (2011), Presidential Early Career Award for Scientists and Engineers (2011), Blavatnik Faculty Award (2013), and ACM-Infosys Foundation Award (2013).

Kristine Yu

Kristine Yu

Department of Linguistics, University of Massachusetts, Amherst
Linguistic tone and the input to computational models of sentence comprehension 
Co-sponsored by Initiative in Cognitive Science
Note different location

Friday, November 13, 2015 • 12:30-p.m.-2:00 p.m. Lunch will be provided, beginning at 12:15
Integrated Learning Center 231

Abstract:  Consider a sentence like “I met the daughter of the colonel who was on the balcony.” When you hear this sentence, there are two possible interpretations: (1) the daughter was the one who was on the balcony or (2) the colonel was the one was on the balcony. These two different interpretations correspond to different syntactic structures of the sentence: that is, the same string of words has two different interpretations because the way the parts of the sentence are related to one another is different in those two interpretations. It has long been known that aspects of how the sentence is spoken, e.g., where the speaker pauses, how the pitch of the speaker’s voice goes up and down, might offer clues to disambiguation in sentence comprehension in a sentence like the example sentence given above. However, the standard approach in computational models of sentence comprehension is to start with an input of a string of words, and to have thrown away the information about how the sentence was uttered. In this talk, I offer perspectives on (1) why clues from the way a sentence was spoken has not been incorporated into computational models of sentence comprehension and (2) why information from the way a sentence was spoken should be incorporated into these computational models, and how we might start tackling this project.

Bio:  My research focuses on speech sounds and how they are patterned in natural language. In particular, I’m interested in aspects of prosody, i.e., (1) tone--the phenomenon of pitch differences distinguishing between word meanings, e.g., in Mandarin, “ma” with a high level pitch means “mother,” but “ma” with a dipping pitch shape means “horse”; and (2) intonation--the phenomenon of pitch differences distinguishing between sentential meanings, e.g., “He ate already” with a rising pitch shape is a question, but “He ate already” with a falling pitch shape is a declarative--and (3) how language structure is chunked in the grammar based on these and related phenomena. My research involves the application of machine learning classification algorithms for understanding how tonal and intonational elements are encoded and decoded in the acoustic speech signal as well as building computational models for recovering prosodic grammatical structure from the incoming speech stream.

Matthew C. Ingram

Matthew C. Ingram

Assistant Professor of Political Science and Research Associate, Center for Social and Demographic Analysis, University at Albany, SUNY
Social Structure, Attitudes, and Law: Network Co-Evolution among Judges in Mexico 
Co-Sponsored with Legal Studies/Political Science
Note different location

Friday, November 6, 2015 • 12:30 p.m.-2:00 p.m.
Thompson 620
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  A broad array of existing research shows that attitudes shape behavior, and this relationship holds for judges and their behavior both on and off the bench. Yet, there is little empirical evidence regarding the content and distribution of judicial attitudes, especially outside the U.S., and even less evidence regarding the manner in which attitudes and other ideational phenomena transfer or diffuse among judges, including the U.S. Building on two survey waves of judges in the Mexican state of Michoacán (2011 and 2015), this paper contributes original data on the attitudes and network ties among legal elites, applying a stochastic actor-oriented model with Simulation Investigation for Empirical Network Analysis (SIENA) to examine the coevolution of attitudes and social structure, i.e., the simultaneous processes of selection and influence. The analysis complements specialized literatures in comparative judicial politics and broader literatures on socialization and policy diffusion, clarifying our understanding of the role of judicial networks in strengthening democracy and the rule of law.

Bio:  Matthew C. Ingram is Assistant Professor of Political Science and Research Associate at the Center for Social and Demographic Analysis at the University at Albany, SUNY. His research is primarily concerned with law and democracy in Latin America, examining the political origins of justice reforms, judicial behavior, and violence, and emphasizing a subnational level of analysis. His academic work has been published in the peer-reviewed journals Comparative Politics, Journal of Law, Economics, and Organization, Government and Opposition, Latin American Politics & Society, and Latin American Research Review, and his book manuscript on state-level judicial reform in Brazil and Mexico is forthcoming with Cambridge University Press (scheduled release November 2016). Ingram holds a law degree (J.D. 2006) and a Ph.D. in political science (2009), both from the University of New Mexico. He was born and raised in Mexico and speaks English, Spanish, and Portuguese.

Albert-László Barabasi

Albert-László Barabasi

Northeastern University
Network Science: From Structure to Control 
Co-Sponsored with the Center for Data Science

Friday, October 30, 2015 • 12:30-p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Systems as diverse as the world wide web, Internet or the cell are described by highly interconnected networks with amazingly complex topology. Recent studies indicate that these networks are the result of self-organizing processes governed by simple but generic laws, resulting in architectural features that makes them much more similar to each other than one would have expected by chance. I will discuss the order characterizing our interconnected world and its implications to network robustness, and control. Indeed, while control theory offers mathematical tools to steer engineered and natural systems towards a desired state, we lack a framework to control complex self-organized systems. I will discuss a recently developed analytical framework to study the controllability of an arbitrary complex directed network, identifying the set of driver nodes whose time-dependent control can guide the system’s dynamics.

Bio:  Albert-László Barabási is the Robert Gray Dodge Professor of Network Science and a Distinguished University Professor at Northeastern University, where he directs the Center for Complex Network Research, and holds appointments in the Departments of Physics and College of Computer and Information Science, as well as in the Department of Medicine at Harvard Medical School and Brigham and Women Hospital in the Channing Division of Network Science, and is a member of the Center for Cancer Systems Biology at Dana Farber Cancer Institute. A Hungarian born native of Transylvania, Romania, he received his Masters in Theoretical Physics at the Eötvös University in Budapest, Hungary and was awarded a Ph.D. three years later at Boston University. Barabási latest book is "Bursts: The Hidden Pattern Behind Everything We Do" (Dutton, 2010) available in five languages. He has also authored "Linked: The New Science of Networks" (Perseus, 2002), currently available in eleven languages, and is the co-editor of "The Structure and Dynamics of Networks" (Princeton, 2005). His work lead to the discovery of scale-free networks in 1999, and proposed the Barabasi-Albert model to explain their widespread emergence in natural, technological and social systems, from the cellular telephone to the WWW or online communities.

Christoph Riedl

Christoph Riedl

Information Systems, Northeastern University, IQSS Harvard University
The Learning Curve of Design Work 

Friday, October 16, 2015 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Much is known about learning curves of production work, characterized by improvements in meeting technical measures of performance. In contrast, little is known about learning in the context of design work, characterized by social measures of performance, even though design work is of increasing interest to organizations. To characterize the design learning curve, we drew on data from over 170,000 design submissions to a online platform hosting design contests over a ten year period. While the design learning curve mirrored production learning curves, it was also characterized by a periods of initial investment prior to realizing returns from experience and in our empirical setting saw a rate of progress of about half of typical production learning curves. We provide results on the role of constraints on the shape of the learning curve, and we discuss the implications of design learning curves for theories of innovation, theories of individual and organizational learning, and for the practice of design.
Co-authored with Victor P. Seidel, Babson College and Harvard SEAS

Bio:  Christoph Riedl is assistant professor for Information Systems at the D'Amore-McKim School of Business at Northeastern University. He also holds an appointment at and the College of Computer & Information Science and is a core faculty at the Network Science Institute. He is a fellow at the Institute for Quantitative Social Science (IQSS) at Harvard University. He is recipient of a Young Investigator Award (YIP) for his work on social networks in collaborative decision-making. Before joining Northeastern University he was a post-doctoral fellow at Harvard Business School and IQSS. He received a PhD in Information Systems from Technische Universität München (TUM), Germany in 2011, a MSc in Information Systems in 2007, and a BSc in Computer Science in 2006. His work has been funded by NSF and published in leading business and computer science journals including Management Science, Communications of the AIS, and International Journal of Electronic Commerce.

Friday, October 9, 2015 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151

Faculty and graduate students come to a research panel and reception to celebrate and welcome 14 new CSSI affiliates (including 8 new hires) who joined CSSI in the past year. Today, six new affiliates will give brief introductions to their research interests, and we will spend the rest of the time sharing lunch, mixing and mingling across colleges. The following affiliates will present:

The following affiliates will present:
Seth Goldman, Communication
Chaitra Gopalappa, Mechanical and Industrial Engineering
Anna Nagurney, Operations and Information Management
Brendan O'Connor, Computer Science
Matthias Steinrücken, Biostatistics and Epidemiology
John Sirard, Kinesiology

Leontine Alkema

Leontine Alkema

Biostatistics and Epidemiology, University of Massachusetts
Statistical Modeling in Global Health: A Selection of Recent Developments and Future Opportunities in Child, Maternal and Reproductive Health 

Friday, October 2, 2015 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151
Lunch will be provided, beginning at 12:00
Talk begins at 12:30

Abstract:  Are developing countries making progress in reducing child and maternal mortality? How do we assess whether girls’ mortality is elevated in countries where a preference for sons may exist? How can countries best monitor unmet need for contraceptive methods, and set targets for improving access?
Questions like these require answers to inform the global health agenda in child, maternal and reproductive health. In this presentation, I will discuss some of the general challenges for answering such questions and give examples of statistical models that we used to provide answers. A Bayesian B-splines regression modelling approach to estimating child mortality and a Bayesian hierarchical time series model for assessing outlying child mortality gender ratios will be discussed in more detail. I will conclude with some future directions for “Stats in action” within global health research.

Bio:  Leontine Alkema is Assistant Professor of Biostatistics at UMass Amherst. Her research focuses on the development of statistical models to assess and interpret demographic and population-level health trends and differentials, generally on a national level, for all countries in the world. She collaborates with various United Nations agencies to make available improved estimation methods and resulting estimates to diverse international audiences

mountain

"Multi-Agent Random Walk" (informal mixer across our 19 departments, with boots on)


Saturday, September 26, 2015 • 10:00 a.m.
Park at the Notch Visitor's Center (1500 West Street, Amherst)

The UMass Computational Social Science Institute invites all to our third Computational Social of 2015-16, a group hike in Amherst. This will be an opportunity to mingle and network with CSSI faculty and grad student affiliates.

Begin at the Notch Visitors Center parking lot, hike westward up Bare Mountain and continue to Mt. Hitchcock on the Metacomet-Monadnock Trail, then return to the Notch. This is <2 hour moderate hike, ~2.5 miles and 1000ft total elevation gain. Bring your own water and trail snacks.

Friday, September 18, 2015 • 12:30 p.m.-2:00 p.m.
Computer Science Building, Room 150/151

Faculty and graduate students come to a research panel and reception to celebrate and welcome 14 new CSSI affiliates (including 8 new hires) who joined CSSI in the past year. Today, six new affiliates will give brief introductions to their research interests, and we will spend the rest of the time sharing lunch, mixing and mingling across colleges. The following affiliates will present:

The following affiliates will present:
Leontine Alkema, Biostatistics & Epidemiology
Ina Ganguli, Economics
Kim Geissler, Public Health Promotion and Policy
Justin Gross, Political Science
Ethan Yang, Civil & Environmental Engineering
Rodrigo Zamith, Journalism

Friday, September 11, 2015 • 5:15 p.m.
Amherst Brewing Company

This is a time for informal conversations and networking among faculty, grad students, and guests of CSSI.

Sid Redner

Sid Redner

Santa Fe Institute
Statistics of Basketball Scoring and Lead Changes 
Colloquium co-sponsored by Physics and CSSI.
Note non-standard date, time, and location

Wednesday, September 9, 2015 • 4:00PM (Refreshments at 3:45)
Hasbrouck 124

Abstract:  Watching basketball is nearly the same as watching repeating coin tossings! By analyzing game-scoring data from recent NBA basketball seasons, scoring during a game is well described by a continuous-time anti-persistent random walk, with essentially no temporal correlations between successive scoring events. As illustrations of this random-walk picture, we show that the distribution of times when the last lead change occurs and the distribution of times when the score difference is maximal are both given by the celebrated arcsine law—a beautiful and surprising property of random walks. We also use the random-walk picture to construct the criterion for when a lead of a specified size is "safe" as a function of the time remaining in the game. The obvious application to game-time betting is left as an exercise for the interested.

mountain

Computational Social


Saturday, September 5, 2015 • 10:00 a.m.
Park at the Notch Visitor's Center (1500 West Street, Amherst)

The UMass Computational Social Science Institute invites you to our first Computational Social of 2015-16, a group hike in Amherst. This will be an opportunity to mingle and network with CSSI faculty and grad student affiliates, and to welcome new students.

Hike up Mt. Norwottuck on the Metacomet-Monadnock Trail, returning on the Robert Frost Trail via the Horse Caves. This is ~2 hour moderate hike, ~4 miles and 850ft elevation gain. Bring your own water and trail snacks.