Tanzeem Choudhury 2002


Principal Investigators: Tanzeem Choudhury and Alex Pentland                   

The first successful experiment in automatically learning the communication patterns

 within a community and the interaction dynamics using wearable sensing devices



The  "Sociometer"





Sensing and Modeling Human Networks

Tanzeem Choudhury

MIT PhD Thesis

August 2003

Modeling Face-to-Face Communication using the Sociometer
Choudhury, T., and Pentland, A
To Appear in: Proceedings of the International Conference on Ubiquitous Computing, Seattle, WA. October 2003 

Sensing and Modeling Human Networks using the Sociometer, Choudhury, T., and Pentland, A
To Appear in: Proceedings of the International Conference on Wearable Computing, White Plains, NY. October 2003 

Learning Communities: Connectivity and Dynamics of Interacting Agents, Choudhury, T., Clarkson, B., Basu, S., and Pentland, A. To appear in the Proceedings of the International Joint Conference on Neural Networks - Special Session on on Autonomous Mental Development. July 2003. 

The Sociometer: A Wearable Device for Understanding Human Networks, Choudhury, T. and Pentland, A. (November 2002) MIT Media Lab TR# 554. Presented at the Conference on Computer Supported Cooperative Work (CSCW '02) (Workshop: Ad hoc Communications and Collaboration in Ubiquitous Computing Environments)

In the Shortcuts project, we are developing methods to automatically and unobtrusively learn the social network structure that arises within a group based on data collected using the sociometer. The questions we are exploring are:

-  Who are the key players in the community? 
-  What are the dynamics of people's interactions and how individual influence each other's   


-  How does information diffuse within the community?


The Experiment:

-  A group of  23 people within a community agree to wear sociometers - 6 hours daily.
-  We collect information about their interactions over  an extended period of time (2weeks).
-  Learn group structure and dynamics using statistical  pattern recognition techniques.
-  Learn how interaction dynamics influence the social network relationships

Experimental Details:

The users agree to have the device on them for the duration that they are on MIT campus. We collected the following data from each individual  -

-  his/her neighborhood information, i.e. the people nearby using IR tags 
-  speech features - spectral features, speech energy, duration, and pitch (using onboard  

-  motion information (using two 2-axis accelerometer).


We have collected 1518 hours of interaction data from 23 individuals over two-weeks.

Hardware Details:

We use the hoarder board with the multi-sensor board extension for data collection in this project. The hoarder board is a microcontroller-based data acquisition platform suitable for user sensing applications. Developed by Vadim Gerasimov, Rich Devaul and Josh Weaver, the general-purpose hoarder board is customized for particular applications through specially designed daughter boards, which provide the sensing hardware.


Special thanks to Brian Clarkson for the effort he put into the design of the shoulder mount.


The use of human subjects for this project has been approved by COUHES (application # 2889)