Integrating Agent Communication Languages in Open Services Architectures

An Open Service Architecture is a software infrastructure that makes an open set of services available to users and agents. The role of agents is to retrieve, execute and compose available services providing more sophisticated instances of them. In order to achieve this goal, agents need to communicate with and use services provided by other agents. Although several Agent Communication Languages (ACLs) have been developed in the last few years providing high level communication primitives, for example based on speech acts, they are still not fully integrated in open service architectures. In this paper we address this issue studying the integration of Agent Communication Languages based on speech acts in these services oriented and geographically distributed architectures. We present the design of a knowledge-level fault tolerant ACL which concerns with the use, request and supply of knowledge, despite possible crashes of remote sites. We provide a formal definition of ACL primitives and a formal specification of the underlying architecture using an algebra of actors. We argue that, to integrate current ACLs in service-oriented geographically distributed architectures, they should be extended with: 1) knowledge-level one-to-many primitives, 2) support for anonymous (contents based) interaction, 3) an hidden unreliable failure detection mechanism. . Dipartimento di Scienze dell’Informazione, University of Bologna, Via Mura Anteo Zamboni, 7, 40127 Bologna, Italy. E-mail: dragoni,gaspari @cs.unibo.it