Reasoning About Communication – A Practical Approach Based on Empirical Semantics

Abstract

Abstract. Given a specification of communication rules in a multiagent system (in the form of protocols, ACL semantics, etc.), the question of how to design appropriate agents that can operate on such a specification is a very important one. In open systems, the problem is complicated even further by the fact that adherence to such a supposedly agreed specification cannot be ensured on the side of other agents. In this paper, we present an architecture for dealing with communication patterns that encompass both a surface structure of admissible message sequences as well as logical constraints for their application. This architecture is based on the InFFrA social reasoning framework and the concept of interaction frames. It assumes an empirical semantics standpoint by which the meaning of communication is pragmatically interpreted through decision-theoretic optimality considerations of a reasoning agent. We introduce the abstract architecture and a formal model and present experimental results from a complex domain to illustrate its usefulness.

Acquiring and adapting probabilistic models of agent conversation

“…This allows for an integration of the results presented here with previous work on interaction frames, particularly an architecture for reasoning about communication within the framework of empirical semantics [5,4] and an application of hierarchical reinforcement learning to the task of learning communication strategies [14,6]. The basic principles of our approach, however, could also be applied to other, possibly more complex, forms of representation.…”

“…Depending on the perspective from which empirical observations are taken, different applications of interaction frames are possible. As shown in [5,4], individual agents can put them into relation to their private goals and use them to derive their communicative actions in order to "communicate optimally" towards these. From the perspective of an external observer, on the other hand, interaction frames can be interpreted as a global model of the communication in a MAS and hence used to measure the performance of the MAS or of individual agents w.r.t.…”

“…Before turning to the acquisition and adaptation of frame-based empirical semantics, we quote [4] for a formal definition of a particular instance of the interaction frame data structure. This definition uses a language M of speech-act [1] like message and action patterns of the form perf(A, B, X) or do(A, Ac).…”

An integrated framework for adaptive reasoning about conversation patterns

“…InFFrA, which is a concrete instance of InFFrA suggested by Fischer and Rovatsos [5] frame trajectories are modelled as linear sequences of message patterns (i.e. speech-act like messages which may contain variables for the participating agents and content elements) which represent discrete, turn-taking communicative encounters between two agents.…”

“…Let s : M * c × 2 L → S be some state abstraction function 5 that returns a state for each pair (w, KB ) of perceived encounter prefix w and belief KB . If we regard each frame F ∈ F as an option in the above sense, we can apply the Semi-MDP Q-learning update rule…”