Semantic social overlay networks

Abstract

Knowledge sharing in a virtual organization requires a knowledge life cycle including knowledge provisioning, terminology alignment, determination of resource location, query routing, and query answering. We focus on determining a relevant resource in a completely decentralized setting such as necessitated by peer-to-peer knowledge management in virtual organizations. Requirements for this task include, e.g., full autonomy of peers as well as full control over own resources and therefore preclude prominent resource location and query routing schemes such asdistributed hash tables. In order to tackle given requirements in this chapter we introduce use a resource location and query routing approach called INGA [23, 24, 31]. It exploits metaphors known from online social networks as well as the semantic similarity between queries and meta data of annotated documents. To adapt to the dynamics of the networks and to bound the local index we present an index update policy combining temporal, semantic and community locality. To further boost performance and enhance recall in a dynamic setting we introduce in INGA recommender and bootstrapping overlays. We have built a network simulator and conducted extensive experiments under realistic conditions. Results show that INGA outperforms other state-of-the-art approaches significantly while it displays small world characteristics.The approach has been fully tested in simulation runs and implemented in the system Bibster (http://bibster.semanticweb.org). © 2010 Springer Science+Business Media, LLC.