Logical Formalization of Concurrent Actions for Multi-Agent Systems

This paper presents a logical approach to formalization of some aspects of multi-agent systems in an incompletely known world. The contribution on this regard is as follows. We formalized reasoning about concurrent actions inside the situation calculus framework which has the suuciently clear and well-understood semantics to express diierent aspects of agent theory. Conclusions can be drawn by default about results of a group of concurrent actions which may include unknown or unspeciied actions. The approach allows to examine all models corresponding both to diier-ent assumptions about possible failures in performing of one or another action, and to assumptions whether a particular action was executed really. This provides an opportunity to predict the consequence of joint activity of agents operating in complex environment and to explain apparently unexpected changes. These results are achieved by choosing a suitable circumscription policy. The approach provides a general and formal framework for design and veriication of wide class of agent-oriented systems such as telecommu-nication systems, network management, and evolving distributed data bases.