A Distributed Symbolic System for AI applications: Design Requirements and Preliminary Results

In this paper we describe a distributed symbolic system that may be viewed as a collection of services, where each one may perform a part of the entire computation. The computational model relies on a symbolic client/server protocol, which uses an application-level mechanism for synchronizing and controlling processes. The intraprocess and interprocess management are obtained using the continuation of a process as the means for context switching and resuming. In this way this system does not rely on any control feature outside the Lisp language. To validate the design criteria of our distributed system, theorem provers seems to be good candidates, since they exploit several methods when searching for a solution. For this reason in this paper we also present the analysis of the requirements for designing a distributed proof generator based on the CMPT, a symbolic Lisp system for automating tutoring for sentential logic, to assess the adequacy of the symbolic distributed system to support the development of real symbolic applications.