Information Retrieval on Text using Concept Similarity

Abstract

Retrieving proper information from internet is a huge task due to the high amount of information available there. Identifying the individual concepts according to the queries is time consuming. To retrieve documents, keyword based retrieval method was used before. Using this type searching, the relationship between associated keywords can’t be identified. If the same concept is described by different keywords, inaccurate and improper results will be retrieved. Concept based retrieval methods are the solution for this scenario. This gives the benefit of getting semantic relationships among concepts in finding relevant documents. Irrelevant documents can be eliminated by detecting conceptual mismatches, which is another benefit obtained from this. The main challenges identified are the ambiguity occurring due to multiple nature of words for the same concepts. Semantic analysis can reveal the conceptual relationships among words in a given document. In this paper the potential of concept-based information access via semantic analysis is explored with the help of a lexical database called WordNet. The mechanism is applied in the selected text documents and extracting the Synonym, Hyponym, Hypernym of each word from WordNet. The ranking will be calculated after checking the frequency rate of each word in the input documents and a hierarchy model will be generated according to the ranking.