Towards Content Aggregation on Knowledge Bases through Graph Clustering.
Christoph Schmitz · 2005
Recently, several research projects such as PADLR and SWAP have developed tools like Edutella or Bibster, which have been targeted at establishing peer-to-peer knowledge management (P2PKM) systems. In such a system, it is necessary for participants to provide brief descriptions of themselves, so that routing algorithms or matchmaking processes can make decisions about which communities peers should belong to, or to which peer a given query should be forwarded. In this talk, I propose the use of graph clustering techniques on knowledge bases for that purpose. After a brief round-trip over an ontology-based P2P knowledge management scenario, I will demonstrate the automatic generation of self-descriptions of peers ’ knowledge bases through the use of graph clustering. Viewing the knowledge base of a peer as a graph consisting of concepts and instances, one can employ clustering techniques to partition it into clusters of similar entities. From each cluster, the centroid can then be selected as a representative. This yields a list of entities giving an aggregated self description of the peer. 1 Ontology-Based P2P Knowledge Management