The research topic "Communication Semantics" at the
AI/Cognition group (Technical University Munich) mainly focuses on adaptive, probabilistic models of agent communication in open systems.
Although several approaches to the semantics of agent communication
have been proposed so far, none of them is really suitable for dealing with
autonomy, which is a decisive property of artificial agents, and is most distinct in
open multiagent systems (e.g. on the internet). As a response to this issue, we have introduced the so-called
Empirical Semantics and
Empirical-Rational Semantics approaches to the semantics and pragmatics of formal agent communication languages.
Our models make use of the
fact that the most general meaning of agent utterances lays in their expectable
consequences in terms of subsequent actions and events, and that communications result from hidden
but nevertheless rational and to some extent reliable agent attitudes. Based on this paradigm, we develop formal frameworks which enable the empirical derivation of
communication meaning from the observation of agent utterances, and
introduce thereby a predictive as well as utility-oriented perspective of communication semantics.
Our research is accompanied with the development of software applications in order to evaluate and apply our theoretical approaches.
We also welcome suggestions from academia and industry for research- and application-oriented cooperations.