Incorporating a Learning Capability into the KBSA.
Abstract
This report describes the research to investigate and demonstrate techniques for incorporating machine learning into the Knowledge-Based Software Assistant (KBSA) program. Advances in failure-based machine learning have made possible the incorporation of these techniques into advanced development paradigms, especially those dependent on acquisition of knowledge as part of the paradigm. The KBSA program, initiated in 1983, is a long-term program to introduce dramatic changes in the very nature of software development. The changes will arise through a new paradigm of development, based on a requirements and design focused development activity. KBSA enables a shift in the leve of development from the lower levels to the highest levels of life cycle through the formalisms used to specify those levels. Changes, to the system are effected through transformations of the requirements or design specifications and implemented through automated mechanisms. This paradigm will dramatically improve both the quality and productivity of the software development process. One of the basic tenets of the program is to automate the assimilation of 'knowledge' into the software assistant. The developer must be able to take advantage of prior events as stored with the KBSA. Knowledge acquisition and the automation of the process of acquiring knowledge are basic to this concept. Machine learning embodies the concept of acquiring knowledge that can be stored and effectively used in the KBSA.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 01, 1996
- Accession Number
- ADA315293
Entities
People
- Sidney Bailin