DefinePK

Abstract

Semantic Search and Information Retrieval forms an integral part of various Search Engines in use. Famous search engines such as, Yahoo, Google, Lycos etc. use the concept of semantic search, where the only comparator for the objects under study is semantic similarity between the objects. The general method involves document-to-document similarity search. This sort of search involves the sequential search of documents one after the other, which involves numerous noise effects. An efficient way of improving this technique is the Latent Semantic Indexing (LSI). LSI maps the words under study on a conceptual space. The conceptual space depends on the queries and the document collection. It uses a mathematical function to figure out the similarity between the words, something called as Singular Value Decomposition. It utilizes the words under study and the ones that are being compared and produces appropriate results. The results obtained are free of semantics like synonymy, polysemy etc. Integrating Word Net, a large lexical database of English language is an efficient way to increase the search result. The word under consideration is linked to the application and the semantic similarities of the word are found out. Documents similar to these similarities are then indexed and listed. The proposed model is tested with standard set of Forum for Information Retrieval (FIRE) documents and a comparison with the term based search has been done.