Start Date
4-1988 8:00 AM
Description
Documentation is an important aspect of software, and issues of project personnel turnover, contractual obligations, and so forth all represent significant concerns. Documentation, in essence, provides understanding. We report on the result of an experiment in automatic software documentation/ explanation in which we input a program (written in REFINE™) and output English describing the functionality of the code. Perceptions on the nature of the documentation/explanation task are presented. Significant issues in the design of such systems (e.g., intended audience, source language, abstraction, concept recognition, etc.) are identified. The characterization of this area as "nontriviaT leads to a research plan to broaden the source of solutions to related areas: machine translation, decompilation, question answering systems, pattern (image) recognition, knowledge representations developed for automatic programming, empirical analysis of software modification strategies, and other topics. Potential applications of this area of research are projected, among them: software reuse, intelligent debugging assistant, software porting aids, maintenance assistant, software restructuring.
Empirical Results in Automatic Software Documentation/ Explanation and a Plan to Increase Its Tractability
Documentation is an important aspect of software, and issues of project personnel turnover, contractual obligations, and so forth all represent significant concerns. Documentation, in essence, provides understanding. We report on the result of an experiment in automatic software documentation/ explanation in which we input a program (written in REFINE™) and output English describing the functionality of the code. Perceptions on the nature of the documentation/explanation task are presented. Significant issues in the design of such systems (e.g., intended audience, source language, abstraction, concept recognition, etc.) are identified. The characterization of this area as "nontriviaT leads to a research plan to broaden the source of solutions to related areas: machine translation, decompilation, question answering systems, pattern (image) recognition, knowledge representations developed for automatic programming, empirical analysis of software modification strategies, and other topics. Potential applications of this area of research are projected, among them: software reuse, intelligent debugging assistant, software porting aids, maintenance assistant, software restructuring.
Comments
No other information or file available for this session.