CONCLASS - A FRAMEWORK FOR REAL-TIME DISTRIBUTED KNOWLEDGE-BASED PROCESSING

We have developed a problem-solving framework, called ConClass, that i s capable of classifying continuous realtime problems dynamically and concurrently on a distributed system. ConClass provides an efficient d evelopment environment for describing and decomposing a classification problem and synthesizing solutions. In ConClass, decomposed concurren t subproblems specified by the application developer effectively corre spond to the actual distributed hardware elements. This scheme is usef ul for designing and implementing efficient distributed processing, ma king it easier to anticipate and evaluate system behavior. The ConClas s system provides an object replication feature that prevents any part icular object from being overloaded. In order to deal with an indeterm inate amount of problem data, ConClass dynamically creates object netw orks that justify hypothesized solutions, and thus achieves a dynamic load distribution. A number of efficient execution mechanisms that man age a variety of asynchronous aspects of distributed processing have b een implemented without using schedulers or synchronization schemes th at are liable to develop bottlenecks. We have confirmed the efficiency of parallel distributed processing and load balancing of ConClass wit h an experimental application.