Show simple item record

dc.contributor.authorPavagada, Ravi
dc.date.accessioned2014-03-04T01:11:30Z
dc.date.available2014-03-04T01:11:30Z
dc.date.issued2006-08
dc.identifier.otherpavagada_ravi_r_200608_ms
dc.identifier.urihttp://purl.galileo.usg.edu/uga_etd/pavagada_ravi_r_200608_ms
dc.identifier.urihttp://hdl.handle.net/10724/23490
dc.description.abstractMost contemporary search engines [8, 17, 41] allow searches on keywords and support direct matching of the keywords with document contents. These search engine return Web pages that contain the search terms by performing the direct or pattern matching of search terms with the page contents. Additionally, the matched search terms might appear in any paragraph of the returned page. Hence, most of these searches return large set of matched Web pages that may or may not be relevant to the context of search. Thus, more often than not the users have to sift through the retrieved pages to find the information they are looking for. In this thesis, we address this problem of search by returning meaningful results that are relevant to the search. We present a prototype search and retrieval system for retrieving information from RDF which represents the knowledge contained in the Web documents. We have addressed the problem of search by returning meaningful results that are relevant to the query. Our proposed system uses the concept of keyword search by extending the concept of keyword search, to ontological classes, literals and relationship. The system processes the entered search terms by matching them to the ontological concepts and relationships. The results returned by our system are either a set of triples or a sub-graph relevant to the query. Our system currently doesn t allow searches on documents, but can be extended to support searches on annotated documents. The key feature of our system is that it exploits relationships in RDF and returns a sub-graph relevant to the query and allows users to enter keywords that are related to the ontological concepts and relationships. We adopt an integrated approach that uses the existing knowledge in the ontology and WordNet [38] along with lexical processing to find related words, unlike other systems that either use WordNet [37] or a domain specific ontology [3, 9, 31] to find related words. Additionally, our system accepts multiple search terms per search, unlike other systems [9, 12, 14, 24] that allows a single search term or literal per search. We compared the precision values of a keyword based retrieval system [8] with that of our system. The comparison indicated that the results returned by our system were very accurate and relevant to the query. On the other hand, the other retrieval system returned many Web pages which weren t relevant to the search.
dc.languageeng
dc.publisheruga
dc.rightspublic
dc.subjectKeyword
dc.subjectSemanticWeb
dc.subjectRDF (Resource Description Framework)
dc.subjectXML,\r ontology
dc.subjectSemantic Information Retrieval
dc.subjectRelationship based retrieval,\r Semantic Document Retrieval
dc.titleSupporting keyword search on semantic web documents
dc.typeThesis
dc.description.degreeMS
dc.description.departmentComputer Science
dc.description.majorComputer Science
dc.description.advisorAmit P. Sheth
dc.description.committeeAmit P. Sheth
dc.description.committeePrashant Doshi
dc.description.committeeJohn A. Miller


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record