Seminars & Colloquia
University of Torino, Italy
"Regulating Agent Interactions in Open Multiagent Systems"
Monday November 22, 2010 10:00 AM
Location: 3211, EB2 NCSU Centennial Campus
(Visitor parking instructions)
Interaction and communication are fundamental abstractions of any distributed system, especially when cross-business and business-to-business systems are to be developed. Multi-agent systems (MAS) are the tools that currently better meet the needs emerging in this context because they offer proper abstractions: in an open MAS the interacting agents are typically designed and implemented by different parties, and may represent conflicting interests.
A key issue in designing MAS is to regulate the interaction between the autonomous parties so that it produces a desirable outcome. To this aim, a particularly interesting approach consists in adopting interaction protocols, meant as shared specifications of behavioral patterns which allow a set of agents to cooperate when they play their respective roles. Besides simplifying the coordination problems, protocols introduce the possibility of performing verification tasks. This aspect is very important because another key concern in this kind of systems is to have guaranties on how the interaction takes place, introducing also a notion of responsibility and of commitment.
Interaction protocols can be formally specified in different ways. Some representations have a procedural nature that captures the allowed interaction flows. Singh and colleagues criticize the use of procedural specifications as being too rigid and propose the more flexible commitment-based approach to protocol specification. The greatest advantage of the commitment-based protocols is that they do not over-constrain the behavior of the agents by imposing an ordering on the execution of the shared actions. Moreover, by giving a shared meaning to the social actions, they make it possible to work on common knowledge, rather than on beliefs about each others' mental state, as instead is done in mentalistic approaches to communication.
Elisa Marengo received her 'Laurea' degree with highest honours in Computer Science in October 2008 at the Universita degli Studi di Torino . Since January 2009 she is Ph.D. student in Computer Science at the Department of Computer Science of the same University, where she won one out of five places in a public competition in October 2008. As Ph.D. student she is currently a member of the research group on Agent-Oriented and Service-Oriented Computing and Knowledge Representation and Reasoning, her advisor is Dr. Matteo Baldoni.
Her interests are related to the representation and reasoning about interaction protocols in MAS and web services and on e-learning and personalization by reasoning in the semantic web.
Host: Munindar Singh, Computer Science, NCSU