An agent-based architecture to support knowledge management in interactive system life-cycle

Several knowledge management issues arise during the life-cycle of an interactive system. During the design phase software engineers need to acquire from users and domain experts knowledge about what they need and about the intended use of the interactive system. Subsequently, due to the phenomenon of user and system co-evolution, users modify their needs and/or their ways of using the system: knowledge about new user habits becomes then a key factor for planning useful software maintenance and extension interventions. In this paper we propose an agent-based architecture to support knowledge management in the complex process of user and system co-evolution. The architecture is based on the visual workshop hierarchy approach, where different kinds of interactive systems are associated in a hierarchical fashion with the different figures involved in system life-cycle. The approach is complemented with a community of agents which may play two kinds of roles: observer agents in charge of observing and recording user-system interaction at each level of the hierarchy, and recognition agents in charge of analyzing observation records and of extracting patterns of interactions to be submitted to the attention of designers at the appropriate level. A case study in earth science is introduced as a concrete application counterpart to the proposed architecture and a sketch about the implementation, which is currently under way, is provided.