Heterogeneous Cooperative Knowledge-Base Systems

Information processing environments have been rapidly changing in many elds: not only changes of network environments and hardware systems, but also enhanced requirements of application systems, database systems, and knowledge-base systems. It is necessary to reconsider how to construct information systems in new environments. For example, in databases, there are many di culties to cope with advanced applications by existing data models such as relational and object-oriented databases. As in scienti c databases, we have to reconstruct fundamental concepts of conventional data models to cope with incomplete and partial objects, heterogeneity of objects, necessity of data mining, schema evolution, and so on. In Japan, the Japanese Fifth Generation Computer Systems (FGCS) project (1982-1993)[2] and its Follow-on project (1993-1995)[3] had accelerated such changes, especially in advanced applications. In the projects, we had confronted difculties in modeling large-scale knowledge information processing applications such as legal reasoning, genetic information processing, and natural language processing. One of the major di culties comes from modeling applications in a single language or a single paradigm. For example, in legal reasoning, we require to embed various features in the system: information retrieval, ordinary database management, logical inference not only with deduction but also with analogy and abduction, and algebraic constraint solvers. How can we model such applications? In our projects, we have designed and developed various powerful languages and systems for knowledge information processing applications[4]. However, each language cannot cover all requirements of the above applications. We started to design such applications based on a multi-agent architecture[5]: a cooperative system consisting of heterogeneous problem solvers, each of which corresponds to a single paradigm. Such systems can work also as a cooperative system consisting of autonomous information systems in distributed environments. Generally speaking, according as expansions of application areas, development of distributed computing environments, and large-scale integration of information systems, a new cooperative framework of heterogeneous problem solvers such as databases, knowledge-bases, constraint solvers, and applications in a distributed environment is indispensable for constructing advanced knowledge information pro-