A KNOWLEDGE REPRESENTATION MODEL FOR MULTIUSER KNOWLEDGE-BASED SYSTEMS

Most existing knowledge-based systems (KBS's) assume a single user or a homogeneous set of users, and thus have a monolithic knowledge base. This paper describes a multiuser knowledge-based system (MKBS) model in which knowledge is represented in a collection of interlinked, largely autonomous user knowledge bases (UKB's), to support use of the system by multiple heterogeneous users. The UKB's are defined such that while each user's view of the knowledge base is limited to their local UKB, problem solving can be achieved using knowledge from multiple UKB's. Different ways of sharing knowledge, including rules, are discussed. A method for rule translation that enables a rule to be used in multiple UKB's is presented. Finally, some issues relating to consistency maintenance and explanation in MKBS are also addressed.