A UNIFIED MODEL FOR SOFTWARE COMPONENT CLASSIFICATION

Abstract

This paper introduces a unified software component classification model. It unifies four widely used retrieval methods, namely faceted methods, free-text methods, enumerative methods, and keyword-based methods, making them special cases. Users can choose and combine the most approprate method or mehtods when searching for components so that the accuracy in component retrieval can be improved. The internal classification structure is decribed. The components in a repository can be retrieval with both query and browse mechanisms. The query algorithm incorporates techniques such as thesauri and degradation functions to maximize the coverage of query results. A simple component identification and version control scheme is also introduced. With this scheme, unique identifiers can be assigned to components withouta cataloging service and component corruptions can be detected easily.