Heterogeneous active agents, III: Polynomially implementable agents

Abstract

In “Heterogeneous active agents, I” (Eiter et al., 1999), two of the authors have introduced techniques to build agents on top of arbitrary data structures, and to “agentize” new/existing programs. They provided a series of successively more sophisticated semantics for such agent systems, and showed that as these semantics become epistemically more desirable, a computational price may need to be paid. In this paper, we identify a class of agents that are called weakly regular—this is done by first identifying a fragment of agent programs (Eiter et al., 1999) called weakly regular agent programs ( WRAPs for short). It is shown that WRAPs are definable via three parameters—checking for a property called “safety”, checking for a property called “conflict-freedom” and checking for a “deontic stratifiability” property. Algorithms for each of these are developed. A weakly regular agent is then defined in terms of these concepts, and a regular agent is one that satisfies an additional boundedness property. We then describe a polynomial algorithm that computes (under suitable assumptions) the reasonable status set semantics of regular agents—this semantics was identified by Eiter et al. (1999) as being epistemically most desirable. Though this semantics is coNP-complete for arbitrary agent programs (Eiter and Subrahmanian, 1999), it is polynomially computable via our algorithm for regular agents. Finally, we describe our implementation architecture and provide details of how we have implemented RAPs, together with experimental results.