|KDnuggets : News : 2009 : n08 : item4||PREVIOUS | NEXT|
FeaturesFrom: Gregory Piatetsky-Shapiro
Subject: SBP09: sociologists and data miners come together to advance social computing
Last month I attended SBP09: Second Workshop on Social Computing, Behavioral Modeling, and Prediction in Phoenix, AZ (many thanks to Prof. Huan Liu, the organizer, for inviting me).
Huan Liu has been working on data and web mining and its applications for many years in particular feature selection and data preprocessing. His recent collaboration with researchers in sociology and cognitive science in social media research converges to social computing and behavioral modeling, an interdisciplinary emerging field with researchers of diverse background but shared interests. His group's recent work in this direction includes two video lectures:
This workshop (SBP'09) was a single-track, two-day event attended by 108 researchers and graduate students.
The workshop was sponsored by NIH, NSF, ONR, AFOSR, and AFRL/II and in-cooperation with ACM SIGKDD.
The goal was to provide a platform for researchers, practitioners, and graduate students from sociology, behavioral science, computer science, psychology, cultural study, information systems, operations research to share, exchange, learn, and develop preliminary results, new concepts, ideas, principles, and methodologies, bridging the gaps between paradigms to encourage interdisciplinary collaborations, advance and deepen our understanding of social and behavioral computing and evaluation in helping critical decision and policy making. The detailed information on the first workshop (SBP’08) (proceedings, presentations and invited speakers) can also be found at www.public.asu.edu/~huanliu/sbp08.
The workshop started with a keynote presentation by Phillip Bonacich (UCLA, Emeritus), about Power and Exploitation in Exchange Networks: A Social-Psychological Model. Prof. Bonacich is one of the founders of the field of social analysis and the inventor of the centrality measure [Bon87]. PageRank, independently invented by Brin and Page used Eigenvector centrality which is mathematically identical to Bonacich measure.
Prof. Bonacich nicely illustrated why network centrality is key to influence by giving an example from the Godfather movie. Godfather, a master of power wanted people to depend on him directly, so his power was measured by number of links to him, or his centrality. He stated
P(AB) = D(BA)
meaning Power of A over B is equal to dependence of B on A.
Mary Lou Maher (NSF) talked about Research Challenges for Computationally Enabled Social and Collective Intelligence. She gave a number of collective intelligence examples, including open source systems, recommender systems, search engines, and Wikipedia.
Her vision was that computers are creating new Human-Computer Intelligent systems: as mediators between people, as tools used by people, or as equal or complementary participants with people. She proposed that we imbue computers with better understanding of people and how we interact both with one another and with computers, at a wide range of granularities.
She saw a key research challenge as "Understanding key properties of emergent intelligence".
William H. Batchelder, a leading expert on psychology and social sciences, talked about Cultural Consensus Theory, which is an approach to pooling information from different sources.
The task is to question people who may share cultural knowledge unknown to the researcher. The researcher can construct questionnaire items but does not know the answers, if any, that best represent the shared knowledge. The researcher also does not know the cultural competence and response bias of each person.
Batchelder showed that a social network model, with a good deal of math, a Bayesian formulation and MCMC methods, can be used to estimate the consensus answers.
Shade Shutters (ASU) talked about Punishment, Rational Expectations, and Relative Payoffs.
He described the modification to the familiar prisoners dilemma, where the rational choice for each prisoner was to cooperate in either. However, the experiments where people played a version of this game showed that the actual behavior of subjects was different. The players were not maximizing their absolute "payoff", but rather the difference between their "payoff" and the other player.
Shutters argued that we should look for evolutionary biology for guidance.
Natural selection is driven by relative fitness. Thus evolutionary biology implied that assumptions of rational choice theory are not correct, and people do not maximize "absolute" payoffs, which are hard to meadure, but relative payoffs.
This is important since international treaties are often designed to convert a dilemma to a "win-win" scenario.
If relative payoffs drive behavior then the treaty may have no effect.
He concluded that behavior in networked populations is better predicted by a biological, relative payoff model, and if relative payoffs drive behavior, restructuring payoffs may not have intended effects.
Many posters were presented during a workshop dinner the first night http://www.public.asu.edu/~huanliu/sbp09/program.html and there was a good mixing of experts from different backgrounds.
On the second workshop day, Alex Pentland from MIT Media Lab gave a keynote talk on Reality Mining: From Profiles and Demographics to Behavior .
Dr. Pentland's work on Reality Mining has attracted great attention, being profiled in New York Times: The Cellphone, Navigating Our Lives (Feb 16, 2009), named 'Breakthrough Idea of 2009' by Harvard Business Review, and 'a technology poised to change the world' by MIT Technology Review.
Pentland's lab invented the technology of reality mining, which analyzes sensor data to extract subtle patterns that predict future human behavior. These predictive patterns begin with biological honest signals, (such as tone of voice, body language), human behaviors that evolved from ancient primate signaling mechanisms, and which are major factors in human decision making in situations ranging from job interviews to first dates.
By using data from mobile phones, electronic ID badges, or digital media to track these honest signals, researchers can create a 'gods eye' view of how the people in organizations interact, and even 'see' the rhythms of interaction for everyone in a city.
In 2006 the MIT researchers were asked to help improve the operations of a large call center. They set up measurement devices which captured not the operator words, but only the physical voice signal: the measured variations in tone and pitch. Even so, Pentland and his researchers predicted accurately, after only a few seconds of listening, the ultimate success or failure of almost every call. Successful operators, it turned out, speak little and listen much.When they do speak, their voices fluctuate strongly in amplitude and pitch, suggesting interest and responsiveness to the customer’s needs.
MIT researchers developed small wearable sensors that measure such "honest signals" and gave them to student volunteers. The results were quite amazing. (..details..)
Sensors and Privacy
Learning can be done effectively from anonymized profiles.
Dr. Pentland company Sense Networks is now commercializing these applications in
Workshop program and slides are available at www.public.asu.edu/~huanliu/sbp09/program.html
[Bon87] Power and centrality: A family of measures, P Bonacich - American Journal of Sociology, 1987, http://www.jstor.org/pss/2780000
|KDnuggets : News : 2009 : n08 : item4||PREVIOUS | NEXT|
Copyright © 2009 KDnuggets. Subscribe to KDnuggets News!