Ontology engineering and routing in distributed knowledge management applications

The dynamics of contemporary business conditions has caused a decentralization of the traditional enterprize organization and a shift towards transferring responsibilities from individuals organized in centralized hierarchical structures to autonomous teams. In particular innovative and value creating knowledge workers involved in research and development, marketing and management activities are affected by this change. This thesis explores the implications this organizational trend has on knowledge management systems and proposes solutions to some of the arising challenges. It introduces a decentralized knowledge management system allowing knowledge workers to share knowledge in a peer-to-peer fashion. Within this system ontologies provide the conceptual backbone to represent and share the knowledge located at individual sites. We investigate such decentralized ontology-based knowledge management systems from an organizational and a technical perspective. In particular the work addresses the building and maintenance of ontologies and the ontology-based routing of queries in this distributed context. DILIGENT proposes an ontology engineering methodology to support the building and evolution of ontologies in decentralized settings. A major concern in such an environment is the construction of consensually shared ontologies to enable knowledge exchange. With this respect the methodology aids ontology users in handling such a shared ontology, while preserving their autonomy by allowing for local variations. It includes an argumentation framework which structures ontology engineering discussions, ensuring the coherence and traceability of the process and facilitating later reuse by explicitly encoding assumptions underlying certain modeling decisions. The DILIGENT methodology is supported by tools and is empirically validated within three case studies. REMINDIN’ is an ontology-based routing algorithm demonstrating the value of shared ontologies for knowledge retrieval in peer-to-peer knowledge management systems. We examine available network topologies with respect to the requirements of a distributed knowledge management scenario, concluding that pure unstructured peer-to-peer networks meet these requirements. REMINDIN’ is a routing algorithm for such network topologies. It exploits social metaphors in conjunction with semantic information in order to select and forward queries to remote peers. We tested the algorithm through simulations resembling real-world ontology-based peer-to-peer systems. This evaluation showed that REMINDIN’ performs better in comparison to related approaches. The thesis demonstrates that ontology-based distributed knowledge management is feasible from an organizational as well as technical point of view. The requirements of the