Research Project

Machine Learning and Intelligent Agents Logo and Link

Machine Learning and
Intelligent Agents

Social Interactions in Multi-Agent Systems


Maria Fasli

Project Description:

Agents rarely act in isolation; in contrast they are increasingly required to act as elements of large and complex systems and cooperate and coordinate with a number of other agents. Reasoning about cooperative activity and teamwork in the context of formal theories requires a very rich ontology of social and collective attitudes such as mutual beliefs and intentions, commitments as well as normative concepts such as obligations and rights.

This project is currently exploring an approach in which the central idea is that stability and regulation of activity within a group of agents can be accounted for by means of a complex web of roles, commitments obligations and rights. In particular, commitments are considered to be the attitudes that hold a group of agents together. In pursuit of their own objectives as well as in order to support their collective commitments, agents adopt roles and undertake social commitments. Being semi-autonomous they may decide to drop their commitments and roles, but they may have to bear the consequences of the other agents' prerogative to exercise their rights.


  Tuesday, 26 October 2004