SOMA

Stochastic Opponent Modeling Agents

Summary

Probabilistic logic programs (PLPs) have been proposed as a paradigm for probabilistic logical reasoning with no independence assumptions. In the Stochastic Opponent Modeling Agents project, we have developed a straightforward variant of PLPs called action-probabilistic logic programs (ap-programs) that can be used to model the behavior of certain socio-cultural-economic groups in different parts of the world. Such ap-programs contain rules that state things like “There is a 50 to 70% probability that group g will take action(s) a when condition C holds.”

In such applications, the problem of interest is that of finding the most probable action (or sets of actions) that the group being modeled might do in a given situation, which corresponds precisely to the problem of finding a “most probable world” (MPW). We have developed several exact and heuristic algorithms for more efficiently solving the MPW problem. Using these methods, the SOMA framework can forecast the most likely behavior of a group in a given situation.

Currently, we have automatically extracted ap-programs for 36 terror groups in the Middle East, looking at 41 strategic actions for each group, such as whether they engaged in bombings, kidnappings, who they target, etc.

Project Lead: Prof. VS Subrahmanian

For additional information, please contact Amy Sliva or Prof. VS Subrahmanian.

Last updated: November 2009 by Amy Sliva.

Project Contributors

Publications

This page will be updated as our work enters print. For information about receiving draft publications, technical reports, and conference presentations, please do not hesitate to contact team members.

The following sections may include links to restricted access material. Please do not hesitate to contact a group member for instructions regarding how to obtain a username and password.

Presentations

The following sections may include links to restricted access material. Please do not hesitate to contact a group member for instructions regarding how to obtain a username and password.

Live Demonstration

The following link will take you to the SOMA Terror Organization Portal (STOP). STOP allows users to browse automatically extracted SOMA models for 36 terror organizations in the Middle East.