A Hybrid Architecture for Web-based Expert Systems

Abstract

A recent technique is to represent the knowledge base of an expert system in XML format. XML parsers are then used to convert XML data into expert system language code. The code is executed or interpreted when providing responses to user queries. Web based expert system (WBES) architectures may be characterized according to where the application knowledge base resides. Applications of both client and server-sided WBES architectures appear in the literature. A hybrid architecture is proposed where servers provide responses to complex queries using server-based processing of code, and clients handle simple queries using data from the XML knowledge base file. That is, both client and server have access to the knowledge base and share the processing of handling user queries. These concepts are illustrated by a comprehensive description of a small passenger information system, deployed in the hybrid web architecture. A server hosts an XML file describing the passenger network, services, stations and connections. This version of the knowledge base is imported by the client generic web page and used to provide a custom-built user interface consisting of entities derived straight from the knowledge base. At the server, the XML format is converted to Prolog code for handling of complex queries.