Knowledge management system framework for collaborative software maintenance environment

Software Maintenance (SM) environment is highly collaborative, complex, knowledge-driven and important. Organizations depend on how quickly, reliably and cost-effective the software could be changed to meet business demands. In a collaborative environment within SM Community of Practice (CoP), users and software maintainers interact on regular basis to share knowledge to resolve software bugs and to extend new enhancements. However, issue such as problem in identifying and sharing knowledge is still regarded as a major challenge. In this study, a Knowledge Management System (KMS) framework for collaborative Software Maintenance (SM) is formulated based on existing components in general frameworks in Knowledge Management (KM), KMS and SM areas. A survey was carried-out to determine important KMS components to the SM CoP and the components are further examined by experts. As a result, a KMS framework consisting of SM Community of Practice, Knowledge Required, SM Process/Activities, Knowledge Management Activities, SM Tools, Automation/Knowledge Discovery Tools, KMS Infrastructure, Collaborative and Soft Influences components was proposed. The KMS framework was then validated vis-a-vis a prototype to verify selected components are indeed useful for SM CoP. The prototype, called Software Maintenance Collaborative Agent Team (SCAT), uses the Multi-Agent System (MAS) technology to automate knowledge storing and sharing during SM processes. A combined ontology has also been developed to link and extend the business domain to SM processes knowledge. To evaluate the KMS framework components, SCAT is implemented, in an in-house SM department and evaluated against another KMS framework represented by an agent-based KMS tool developed by Rodriguez et al. A survey (based on Wu and Wang KMS Success Model) was conducted to gauge the benefits of both KMS tools, which covers knowledge or information quality and perceived KMS benefits. Based on t-test comparison on overall Rasch estimated measures, it was concluded that SCAT KMS is significantly better than Rodriguez KMS in the above areas. In addition, based on Common Person Equating plot and Measure Differences plots, SCAT is significantly better in important aspects such as complete knowledge portal, sharing specific knowledge and improves quality. In future, SCAT could be enhanced to improve the weak areas