Using version spaces to support incremental searches in a software library

Information spaces of software component descriptions in software libraries tend to be unstructured and ill-defined. Powerful browsing facilities are important for the successful use of software libraries to support software reuse. Browsing is an activity which iterates two processes: search result investigation and search tactic execution. Query refinement has been identified as a common search tactic used by humans. This paper describes an experimental software library system AIRS which exploits the use of some AI techniques to support incremental searching, a type of query refinement. Specifically, AIRS uses the concept of subsumption to organize software descriptions into a generalization hierarchy, and uses version spaces, a technique originated from machine learning, to support incremental searching through the generalization hierarchy