Methodologies for Knowledge-Based Software Engineering

As the science of knowledge representation and automated reasoning advances, AI has the potential to radically change the artifacts, methodologies, and life cycles of software engineering. The most significant change will be when problems are formalized at the level of specifications rather than programs. This will greatly facilitate software reuse and modification. Achieving this potential requires overcoming many technical challenges, particularly the semi-automated synthesis of efficient and correct programs from specifications. The first part of this paper describes several methodologies for program synthesis and compares their ability to control the combinatorial explosion inherent in automated reasonAs knowledge-based software engineering matures and increasingly automates the softwareengineering l i e cycle, software engineering resources will shift toward knowledge acquisition and the automated reuse of expert knowledge for developing software artifxts. The second part of this paper describes methodologies for expanding the software life cycle to the knowledge life cycle. ing.