An association-;based management of reusable software components

Abstract

Processes to use environments which store reusable software components can be classified into “registration” (representation) and “retrieval” (remembering) processes. A conceptual space called “reuse space” is introduced which consists of the presentations of software entities and predicates to define the properties which the target entity should satisfy. The predicate parts are implemented by property definitions for entities, described with the language called HSML, and associative networks. The associative network is structured with using a psychological principle called category-;based induction. In the registration processes, nodes and links, which represent the new entity and the relationships with existing nodes, can be added to the associative networks. In the retrieval processes, the target entity can be remembered by searching the highest rating cluster in the associative networks with the aid of an inference engine. Clustering is performed with the use of coverages and proximities attached to the links in the network. The environment called MANDALA consists of user interfaces for displaying the reuse space on the client stations, a central web server and many distributed local servers which mount the contents of reusable components.