Discovering semantic aspects of socially constructed knowledge hierarchy to boost the relevance of Web searching

Abstract

The research intends to boost the relevance of Web search results by classifyingWebsnippet into socially constructed hierarchical search concepts, such as the mostcomprehensive human edited knowledge structure, the Open Directory Project (ODP). Thesemantic aspects of the search concepts (categories) in the socially constructed hierarchicalknowledge repositories are extracted from the associated textual information contributed bysocieties. The textual information is explored and analyzed to construct a category-documentset, which is subsequently employed to represent the semantics of the socially constructedsearch concepts. Simple API for XML (SAX), a component of JAXP (Java API for XMLProcessing) is utilized to read in and analyze the two RDF format ODP data files, structure.rdfand content.rdf. kNN, which is trained by the constructed category-document set, is used tocategorized the Web search results. The categorized Web search results are then ontologicallyfiltered based on the interactions of Web information seekers. Initial experimental resultsdemonstrate that the proposed approach can improve precision by 23.5%.