Title:
Digital Media and The Relational Revolution in Social Science
View slides
Social science paradigms are invariably grounded in the available methods of data collection. Beginning with administrative records in the late 19th Century, social scientists have collected stores of data on individual attributes, using surveys and records kept by governments and employers. Individual-level data is also aggregated as population statistics for groups of varying size, from households to nation states, and these data are analyzed using multivariate linear models that require the implausible assumption that the observations are independent, as if each respondent was the sole resident of a small island. In comparison, until recently, we have had very limited data about the interactions between people – such as influence, sanctioning, exchange, trust, attraction, avoidance, and imitation. Yet social relations and interactions are the foundation of social life.
The entities that we most need to learn about are the things about which we know the least. The reason is simple: It is much easier to observe friends than to observe a friendship. Social interactions are fleeting and mostly private – one needs to be present at precisely the right moment. Moreover, relations are tedious and error-prone to hand-code and record, given the nuances of interaction, the need for repeated observations as relations unfold over time, and the rapid increase in the number of relations as the size of the group increases. As a consequence, studies of social interactions tend to be static, limited to the structures of interaction without regard to content, and based on very small groups. That is why social science has generally been limited mainly to the study of individuals with individual data aggregated for groups and populations. Except in very small groups, social relations have been just too hard to observe.
All this is rapidly changing as human interactions move increasingly online. Interactions that for the history of humankind have been private and ephemeral in nature now leave a silicon record – literally footprints in the sand – in the form of publicly available digital records that allow automatic data collection on an unprecedented scale. However, social scientists have been reluctant to embrace the study of what is often characterized as the “virtual world,” as if human interaction somehow becomes metaphysical the moment it is mediated by information technologies. While great care must be exercised in generalizing to the offline world, the digital traces of computer-mediated interactions are unique in human history, providing an exceptional opportunity for research on the dynamics of social interaction, in which individuals influence selected others in response to the influences they receive. In my presentation, I will survey recent studies using digital records of interpersonal interaction to address questions ranging from social inequality to diurnal and seasonal mood changes to the spread of protest in the Arab Spring, including contributions by Rob Claxton, Nathan Eagle, Scott Golder, Jon Kleinberg, Noona Oh, Patrick Park, Michael Siemens, Silvana Toska, and Shaomei Wu.
Professor Michael Macy is Goldwin Smith Professor of Sociology and Director of the Social Dynamics Laboratory at Cornell. With support from the U.S. National Science Foundation, he uses computational models, online laboratory experiments, and digital traces of device-mediated interaction to study how cooperation, trust, social norms, and innovations emerge and spread through processes of social influence on dynamic social networks.
Macy pioneered the use of agent-based computational models in sociology, as well as the use of massive data collected from online social networks. From 2005 to 2008, he led a Cornell initiative to promote cross-disciplinary collaborative research on social and information networks. His research has been published in leading journals, including Science, PNAS, American Journal of Sociology, American Sociological Review, and Annual Review of Sociology.
Title:
Using Web Science to Understand and Enable 21st Century Multidimensional Networks
View slides
Recent advances in Web Science provide comprehensive digital traces of social actions, interactions, and transactions. These data provide an unprecedented exploratorium to model the socio-technical motivations for creating, maintaining, dissolving, and reconstituting multidimensional social networks. Multidimensional networks include multiple types of nodes (people, documents, datasets, tags, etc.) and multiple types of relationships (co-authorship, citation, web links, etc).
Using examples from research in a wide range of activities such as disaster response, public health and massively multiplayer online games, Contractor will argue that Web Science serves as the foundation for the development of social network theories and methods to help advance our ability to understand and enable multidimensional networks.
Professsor Noshir Contractor is the Jane S. & William J. White Professor of Behavioral Sciences in the McCormick School of Engineering & Applied Science, the School of Communication and the Kellogg School of Management at Northwestern University, USA. He is the Director of the Science of Networks in Communities (SONIC) Research Group at Northwestern University.
He is investigating factors that lead to the formation, maintenance, and dissolution of dynamically linked social and knowledge networks in a wide variety of contexts including communities of practice in business, translational science and engineering communities, public health networks and virtual worlds. His research program has been funded continuously for over a decade by major grants from the U.S. National Science Foundation with additional current funding from the U.S. National Institutes of Health (NIH), Air Force Research Lab, Army Research Institute, Army Research Laboratory and the MacArthur Foundation.
Title:
Data Mining as a Key Enabler of Computational Social Science
View slides
Observation and analysis of a phenomenon at unprecedented levels of granularity not only furthers our understanding of it, but also transforms the way it is studied. For instance, invention of gene-sequencing and computational analysis transformed the life sciences, creating fields of inquiry such as genomics, proteomics, etc.; and the Hubble space telescope has furthered the ability of humanity to look much farther beyond what we could otherwise.
With the mass adoption of the Internet in our daily lives, and the ability to capture high resolution data on its use, we are at the threshold of a fundamental shift not only in our understanding of the social and behavioral sciences (i.e. psychology, sociology, and economics), but also the ways in which we study them. Massively Multiplayer Online Games (MMOGs) and Virtual Worlds (VWs) have become increasingly popular and have communities comprising tens of millions. They serve as unprecedented tools to theorize and empirically model the social and behavioral dynamics of individuals, groups, and networks within large communities.
The preceding observation has led to a number of multi-disciplinary projects, involving teams of behavioral scientists and computational scientists, working together to develop novel methods and tools to explore the current limits of behavioral sciences. This talk consists of four parts. First, we describe findings from the Virtual World Exploratorium; a multi-institutional, multi-disciplinary project which uses data from commercial MMOGs and VWs to study many fields of social science, including sociology, social psychology, organization theory, group dynamics, macro-economics, etc. Results from investigations into various behavioral sciences will be presented.
Second, we provide a survey of new approaches for behavioral informatics that are being developed by multi-disciplinary teams, and their successes. We will also introduce novel tools and techniques that are being used and/or developed as part of this research. Third, we will discuss some novel applications that are not yet there, but are just around the corner, and their associated research issues. Finally, we present commercial applications of Game Analytics research, based on our experiences with a startup company that we've created.
Professor Jaideep Srivastava is Professor of Computer Science & Engineering at the University of Minnesota, where he directs a laboratory focusing on research in Web Mining, Social Media Analytics, and Health Analytics. He has authored over 275 papers, and his research has been supported by government agencies, including NSF, NASA, ARDA, IARPA, NIH, CDC, US Army, US Air Force, and MNDoT; and industries, including IBM, United Technologies, Eaton, Honeywell, Cargill, and Huawei Telecom. He has an active collaboration with Allina's Center for Healthcare Innovation, where he is a Distinguished Fellow.
He is on the oversight committee for healthcare information architecture for the University of Minnesota. Dr. Srivastava has significant experience inthe industry, in both consulting and executive roles. He has lead a data mining team at Amazon.com, and built a data analytics department at Yodlee. He has provided technology and strategy advice to Cargill, United Technologies, IBM, Honeywell, KPMG, 3M, TCS, and Eaton, and has served as Advisor to the State Government of Minnesota and the Government of India.
He holds distinguished professorships at Heilongjiang University and Wuhan University, China. Dr. Srivastava has BS from Indian Institute of Technology (IIT), Kanpur, India, and MS and PhD from University of California, Berkeley. He is a Fellow of the Institute of Electrical and Electronics Engineers (IEEE), and has been an IEEE Distinguished Visitor. He has given over 150 invited talks in over 30 countries, including more than a dozen keynote addresses at major conferences. He is a co-founder and President of Ninja Metrics, a social analytics company, whose goal is bring his research in this area into the commercial domain.
Title:
Predicting Market Movements: From Breaking News to Emerging Social Media
In this talk I will present seveal studies conducted at the AI Lab of the University of Arizona that aim to understand and predict market movements using text mining, breaking news, and social media.
In “User-Generated Content on Social Media: Predicting New Product Market Success from Online Word-of-Mouth,” we explore the predictive validity of various text and sentiment measures of online WOM for the market success of new products. The context of our study is the Hollywood movie industry where the forecast of movie sales is highly challenging and has started to incorporate online WOM. We first examine the evolvement patterns of online WOM over time, followed by correlation analysis of how various sentiment measures are related to the metrics of new product success. Overall, the number of WOM messages was found to be the most useful predictor of the five new product metrics.
In “AZ SmartStock: Stock Prediction with Targeted Sentiment and Life Support,” we develop a text-based stock prediction engine with targeted sentiment and life support considerations in a real world financial setting. We focus on inter-day trading experiments, with the 5-, 10-, 20-, and 40-day trading windows. We focus on S&P 500 firms in order to minimize the potential illiquid problem associated with thinly traded stocks. News articles from major newswires were extracted from Yahoo! Finance. Life support of a company is extracted from aggregated energy (novelty) of terms used in the news articles where the company is mentioned. The combined Life-Support model was shown to out-perform other models in the 10-day trading window setting.
In “A Stakeholder Approach to Stock Prediction using Finance Social Media,” we utilize firm-related finance web forum discussions for the prediction of stock return and trading of firm stock. Considering forum participants uniformly as shareholders of the firm, suggested by prior studies, and extracting forum-level measures provided little improvement over the baseline set of fundamental and technician variables. Recognizing the true diversity among forum participants, segmenting them into stakeholder groups based upon their interactions in the forum social network and assessing them independently, refined the measures extracted from the forum and improved stock return prediction. The superior performance of the stakeholder-level model represented a statistically significant improvement over the baseline in directional accuracy, and provided an annual return of 44% in simulated trading of firm stock.
Professor Hsinchun Chen is McClelland Professor of Management Information Systems at the University of Arizona. He received the B.S. degree from the National Chiao-Tung University in Taiwan, the MBA degree from SUNY Buffalo, and the Ph.D. degree in Information Systems from the New York University. Dr. Chen had served as a Scientific Counselor/Advisor of the National Library of Medicine (USA), Academia Sinica (Taiwan), and National Library of China (China).
Dr. Chen is a Fellow of IEEE and AAAS. He received the IEEE Computer Society 2006 Technical Achievement Award and the INFORMS Design Science Award in 2008. He has the h-index of 56. He is author/editor of 20 books, 25 book chapters, 210 SCI journal articles, and 140 refereed conference articles covering Web computing, search engines, digital library, intelligence analysis, biomedical informatics, data/text/web mining, and knowledge management.
Title:
Learning Information Diffusion Models from Observation and Its Application to Behavior Analysis
View slides
We investigate how well different information diffusion models can explain observation data by learning their parameters and discuss which model is more appropriate to which topic. We use two models, one from push type diffusion (AsIC) and the other from pull type diffusion (AsLT), both of which are extended versions of the well known Independent Cascade (IC) and the Linear Threshold (LT) models that incorporate asynchronous time delay. The model parameters are learned by maximizing the likelihood of the observed data being generated by an EM like iterative search, and the model selection is performed by choosing the one with better predictive power.
We first show by using four real networks that the proposed learning algorithm correctly learns the model parameters both accurately and stably, and the proposed selection method identifies the correct diffusion model from which the data are generated. We then apply these methods to behavioral analysis of topic propagation using a real blog diffusion sequence, and show that although the inferred relative diffusion speed and range for each topic is rather insensitive to the model selected, there is a clear indication of which topic to better follow which model. The correspondence between the topic and the model selected is indeed interpretable.
Professor Hiroshi Motoda is a professor emeritus of Osaka University and a scientific advisor of AFOSR/AOARD (Asian Office of Aerospace Research and Development, Air Force Office of Scientific Research, US Air Force Research Laboratory). His research interests include information diffusion, data mining, machine learning, knowledge acquisition, scientific knowledge discovery and artificial intelligence in general. He received his Bs, Ms and PhD degrees all in nuclear engineering from the University of Tokyo.
He is a member of the steering committee of PAKDD, PRICAI, DS and ACML. He received the best paper awards from Atomic Energy Society of Japan (1977, 1984) and from Japanese Society of Artificial Intelligence (1989, 1992, 2001), the outstanding achievement awards from JSAI (2000), the distinguished contribution award from PAKDD (2006), Okawa Publication Prize from Okawa Foundation (2007) and outstanding contribution award from Web Intelligence Consortium (2008). He is a fellow of JSAI.
Title:
Analysis of Twitter Unfollow: How Often do People Unfollow in Twitter and Why?
View slides
Unfollow in Twitter offers a unique opportunity to researchers to study the dissolution of relationship. We collected daily snapshots of follow relationship of 1.2 million Korean-speaking users for 51 days and their all tweets. From careful statistical analysis, we confirm that unfollow is prevalent and irrelevant to the volume of interaction. We find that other factors such as link reciprocity, tweet burstiness and informativeness are crucial for unfollow decision. We conduct interview with 22 users to supplement the results and figure out motivations behind unfollow behavior.
From those quantitative and qualitative research we draw significant implications in both theory and practice. Then we use a multiple logistic regression model to analyze the impacts of the structural and interactional properties on unfollow in Twitter. Our model with 42 dependent variables demonstrates that both structural and interactional properties are important to explain the unfollow behavior. Our findings are consistent with previous literature about multiple dimensions of tie strength in sociology but also add unique aspects of unfollow decision that people appreciate receiving attention rather than giving.
Professor Sue B. Moon received her B.S. and M.S. from Seoul National University, Seoul, Korea, in 1988 and 1990, respectively, all in computer engineering. She received a Ph.D. degree in computer science from the University of Massachusetts at Amherst in 2000. From 1999 to 2003, she worked in the IPMON project at Sprint ATL in Burlingame, California. In August of 2003, she joined KAIST and now teaches in Daejeon, Korea.
She has served as TPC co-chair for APSys 2011, ACM SIGCOMM 2007 MobiArch Workshop, and ACM Multimedia 2004, general chair for PAM, and TPC for many conferences, including SIGCOMM 2010, NSDI 2008 and 2010, WWW 2007-2009 and 2011-2012, INFOCOM 2004-2006, and IMC 2009.
Title:
Understanding Player Behaviors from Real-Time Strategy (RTS) Game Data: A Starcraft II case study
In this talk, I will share some of the ongoing efforts in our project to analyze Starcraft II's gameplay data. Specifically, our research goal is to learn how certain types of gameplay segments are associated with player's intents and strategies, how to identify these types of gameplay behaviors automatically from in-game data using data mining and machine learning techniques, how crowdsourcing can help us make sense of a large amount of gameplay data, etc.
Dr. Achananuparp is currently a Research Scientist at Living Anayltics Research Centre, Singapore Management University. Previously, I was a Postdoctoral Fellow working with Professor Ee-Peng Lim in his Social Network Mining group at SMU. His current research interests include social network analysis and mining, behavior modeling, game mining, and text mining.
Prior to coming to Singapore, he recieved his doctoral degree in Information Science at The iSchool at Drexel Univeristy, Philadelphia, under the supervision of Professor Xiaohua Hu. The title of my doctoral thesis is "Similarity Measures and Diversity Rankings for Query-Focused Sentence Extraction".
Title:
Defining the Word “Social” in Games
Is “Social community/activities/interactions” the next big wave in the game market trend? Has Facebook change the way we play game nowadays? These are just some of the many questions that I will be covering along with a brief analysis on various ways of how social interactions can be integrated with the interactive experience from the game medium. I will also offer some outtakes on the current game market trend and provide some insights about the future of the industry as a whole.
Dominic Chai graduated from Computer Engineering at Nanyang Technological University. He has been eager to explore the 'wonders' of the Game Industry. Equipped with 4 years of Game Development experience and a strong passion for games, he has hoped to see his retirement in this industry some day. Dominic participated in the GAMBIT summer internship 2007 as a Scrum-Master to develop a game for two months.
Upon his return, he continued his journey and began to work as a Production Assistant at Mikoishi Pte Ltd, a Singapore games development studio. He is currently an Assistant Producer at the Singapore-MIT GAMBIT Game Lab (Singapore lab), and hopes to learn and work with the professionals and talents the industry has to offer.
Title:
Analyzing 'Fun' in Social Games
In this talk we will investigate what makes games "fun" by introducing the "6-11 Framework", an analysis model successfully taught in game design classes at DigiPen Institute of Technology - Singapore, and use it to gain insights on how games in general and, more specifically, modern social games, deliver engaging and immersive experiences.
Once a proper game analysis methodology has been established, we will proceed to discuss different player typologies and map them into the proposed model to gain a better understanding of how different people can enjoy the same game for different reasons thanks to the broad range of game dynamics available within social and online games.
Professor Roberto Dillon was born in Genoa, Italy and holds a Ph.D. degree in Electrical and Computer Engineering from the University of Genoa. After having worked both in the software/multimedia development industry and in prestigious academic institutions across Europe and Asia, including KTH in Stockholm and NTU in Singapore, he joined the Singapore campus of the DigiPen Institute of Technology where he is currently Chair of the Game Software Design and Production Department, teaching a variety of subjects including Game Mechanics and Game History.
Roberto has led high profile research projects on innovative game mechanics, designed serious and experimental games that raised the interest of the international press and were showcased at events like the "Sense of Wonder Night" in Tokyo and "FILE Games" in Rio de Janeiro and has written two books, "On the Way to Fun" and "The Golden Age of Video Games", published by AKPeters and CRC Press.
