This Small Business Innovation Research Phase I project will investigate the feasibility of providing a natural language processing (NLP) interface to software documentation. The creation, use, maintenance, and re-engineering of sophisticated software systems is limited, in part, by the ability of computer personnel to learn and remember the details of these software systems. This proposal presents a three-pronged approach to reducing this limitation through software documentation. A statistical language system will allow the extraction of design information and user documentation by NLP queries, a syntax-based scanner for the target programming language will allow browsing of the objects in the source code, and a NLP dialog interface will facilitate the entry of program comments. An existing software application will be chosen in the Phase I for use in developing a prototype of the proposed documentation system. Where possible, existing NLP technology will be adapted to support the retrieval of software documentation and the entry of program comments. This research has potential applications in improving the productivity of those engaged in designing, developing, using, maintaining, and re-engineering software.
