Semantic Search in Peer-to-Peer based Digital Libraries
Final doctoral research paper presented at JCDL 2005 Inaugural Doctoral Consortium, 7 June 2005, Denver, Colorado, USA.
This paper gives an outline of a PhD thesis inspired by the information searching over heterogeneous metadata records within the scenario of Peer-to-Peer (P2P) based digital libraries. We survey a representative set of P2P systems in order to build prospective platform and discuss why a super-peer (peer community) based topology is suitable for schema-based P2P applications. Furthermore, we exploit ontologies for storing metadata records in a structured and formal way, and for broadening queries with closely related terms, thus yielding more relevant results. All metadata records into RDF format and a RDF database, such as Sesame, is adopted to store and query these records and schema information. Additionally, since words generally have more than one sense, we have to choose the right sense and highly related terms for extending the query. Such information can be delivered by applying a lexical database, like WordNet, which models terms with their meanings and relationships in between. The resulting prototype system is expected to serve as a framework for re-processing metadata records into structured and semantic-enriched collection in specific domains, as a platform for the searching over heterogeneous collections, and as a source of inspiration for exploiting ontologies in information searching.