Improving Web Search Results for Homonymsby Suggesting Compl

? Springer-Verlag Berlin Heidelberg 2010Improving Web Search Results for Homonyms bySuggesting Completions from an OntologyTian Tian1 , James Geller 1 , and Soon Ae Chun2 1 New Jersey Institue of Technology,Newark, NJ, USA{tt25,geller}@njit.edu2 CSI, City University of New YorkStaten Island, NY, USA{soon.chun}@Abstract. The terms that users pass to search engines are often ambiguous, byreferring to homonyms. The search results in these cases are a mixture of linksto documents that refer to different meanings of the search terms. Currentsearch engines provide suggested query completion terms as a dropdown list.However, such lists are not well organized, mixing completions for differentmeanings. In addition, the suggested search phrases are not discriminatingenough. We propose an approach to suggesting well organized and visuallyseparated search completions based on ontologies. In addition, our approachsupports the use of negative terms to disambiguate the suggested completions inthe list. We present an algorithm to generate the suggested search completionterms. We have developed musician and basketball player ontologies and anOntology-Supported Web Search (OSWS) System for “famous people” thatgenerates the suggested search term completions, based on these ontologies.Keywords: Ontology-supported Web search, semantic search methods,homonyms, negative search terms, suggested completions.1IntroductionUsers’ information needs in the digital era can be fulfilled by keyword-based searchengines. Such search engines have become the universal catalogs for world-wideresources. Unlike the old library catalogs that are mostly searchable by fixed fields(e.g., by authors, titles, and keywords predefined by authors), modern Web search