Integrating information retrieval and domain specific approaches for browsing and retrieval in object-oriented class libraries

Abstract

New techniques for browsing amongst functionally related classes, and retrieving classes from objectoriented class libraries are presented. These techniques make use of two potent, and rea.dily available sources of information: the source code of each class, and its associated documentation. We describe how the integration of information retrieval techniques based on document analysis, and domain specific approaches based on code analysis, permits the construction of class retrieval tools based on natural language queries, and new kinds of browsing tools based on class functionality rather than inheritance. The chief advantages of this approach, compared to previous approaches, are that information about classes is acquired automatically and cheaply - no human intervention is required t,o acquire information about each class; and it is readily scalable and extensible - classes can be added to the library with very little effort. © 1991, ACM. All rights reserved.