Improving the Relevance of Search Results

Synonymy & polysemy of natural languages together with information overload are two main factors that affect the relevance of Web hits. When users submit a query search engines usually return a long list of hits with syntactic similarity. Users are confronted with choosing a needle from a haystack - relevant items from long lists of hits. This book proposes an improved strategy for increasing the relevance of Web search results via search term disambiguation and ontological filtering. Results are classified into an ontology such as Open Directory Project. Semantic characteristics of ontology categories are represented by a category-document and similarities of this and search results are evaluated using a Vector Space Model. Users choose a category to obtain only the search results classified under the selected category. Experimental data show the approach boosts the Web hits precision by more than 20%. The book should help shed some light on Web searching and word sense disambiguation and should be useful to students and researchers in the fields of information retrieval text classification and data mining; or anyone else interested in Web searching.