Case based software maintenance supporting expert system for COBOL domain

Abstract

Software maintenance is the most expensive phase in the software life cycle. Enhancement of productivity is one of the important issues in the software manitenance phase. For the reason, we suggest a case based methodology for COBOL language domain to recommend the maintainable targets and to show the IS alternation history at the level of source code to recall the historical success stories. We develop the case based reasoning and retrieval system for software maintenance with application domain, KEPCO (the Korea Electric Power Corporation). In the system, we represent the architecture of a case which is defined as one event of software maintenance. With the case architecture, we index the cases and build the case-base. The procedures for case based software manitenance supporting is composed of the following steps : 1) Requirement input with six unit requirement types and its related keyword, 2) Case retrieval and reasoning with perfect and partial matching algorithm, 3) Step recommendation with the output part of retrieved cases, 4) Post processing which consisted with filtering out and case storage. A prototype case based software maintenance supporting expert system is implemented as a part of SWMES (Software Maintenance Expert System) using a expert system development tool UNiK on WINDOWS 3.1 environment of 486 PC machine.