Retro grouping University of Washington Computer Science & Engineering
 Sensing and Modeling Dynamic Social Networks
  CSE Home   About Us    Search    Contact Info 

CSE Faculty
 Tanzeem Choudhury
 Jeff Bilmes
Sociology Faculty
 James Kitts
Grad Students
 Danny Wyatt


The structure and dynamics of social networks are of critical importance to many social phenomena, ranging from organizational efficiency to the spread of knowledge and disease. The study of such networks has spawned numerous research conferences and even a series of rapidly growing start-up companies providing software tools for social networking. Academic research in social networks has an abundance of interesting and important questions, but has been faced with a paucity of data rich enough to answer many of these questions. Using a rich new data source, we consider some questions on the dynamics of social networks.

  • How do social networks change over time, over short (weeks) and long (years) time scales?
  • Can one predict shortening paths in the network, and/or predict when the network is likely to break into disconnected components?
  • What is the spatio-temporal distribution of interactions across locations and classes of locations?
  • How do locations serve as hubs or bridges, channeling the evolution of interpersonal networks?
  • Can one predict the rise and fall of popularity of particular locations as hubs for social networks?

We take innovative approach to learning the structure and dynamics of social networks: using sensing and communications tools together with artificial intelligence techniques to unobtrusively study large populations of interacting humans over extended periods of time. This approach will addresses questions applicable to research ranging from ubiquitous computing to machine learning and perception to human-computer interaction.

  • How can we effectively combine data from multiple sensors in order to detect face-to-face human interaction?
  • Can we automatically learn models of interaction dynamics?
  • Can we infer higher-level features of interactions, e.g. social roles, status relations, or emotional connections?

Answering these questions will have significant impacts, not only on the computer science and social science research literature, but also by adding value to applied endeavors such as designing public spaces and office environments, and developing computer collaboration tools.


Towards the Automated Social Analysis of Situated Speech Data
Danny Wyatt, Tanzeem Choudhury, Jeff Bilmes, and James Kitts
Proceedings of UbiComp 08. Seoul, Korea. September 2008.

Learning Hidden Curved Exponential Random Graph Models to Infer Face-to-Face Interaction Networks from Situated Speech Data
Danny Wyatt, Tanzeem Choudhury, and Jeff Bilmes
Proceedings of AAAI-08. Chicago, Illinois. July 2008.

Conversation Detection and Speaker Segmentation in Privacy-Sensitive Situated Speech Data
Danny Wyatt, Tanzeem Choudhury, and Jeff Bilmes
Proceedings of Interspeech 2007. Antwerp, Belgium. August 2007.

Capturing Spontaneous Conversation and Social Dynamics: A Privacy Sensitive Data Collection Effort
Danny Wyatt, Tanzeem Choudhury, and Henry Kautz
Proceedings of ICASSP-07. Honolulu, Hawaii. April 2007.

A Privacy-Sensitive Approach to Modeling Multi-Person Conversations
Danny Wyatt, Tanzeem Choudhury, Jeff Bilmes, and Henry Kautz.
Proceedings of IJCAI-07. Hyderabad, India. January 2007.

Towards Activity Databases: Using Sensors and Statistical Models to Summarize People's Lives.
Tanzeem Choudhury, Matthai Philipose, Danny Wyatt, and Jonathan Lester
IEEE Data Engineering Bulletin 29(1), pp. 49-56. March 2006.

Contact Info

For more information, please contact Danny Wyatt:


CSE logo Computer Science & Engineering
University of Washington
Box 352350
Seattle, WA  98195-2350
(206) 543-1695 voice, (206) 543-2969 FAX
[comments to Danny Wyatt]