Title: Concept Based Information Retrieval from Text Documents
Abstract: This research is intended to develop a concept based information retrieval system for text documents in two phases: Therefore, this idea motivated us to develop a concept based information retrieval system for text documents.This system tries to provide additional semantics as conceptually related words with the help of glosses to the query words and keywords in the documents by disambiguating their meanings.Here, various senses provided by WSD algorithm have been used as semantics for indexing the documents to aid the information retrieval system.Later, this research has also been motivated to do ontology based information retrieval from Tamil text documents which improve the retrieval performance in a better way due to the incorporation of domain semantics.Here, the performance of IR has been improved by including more indexing information about the documents such as associated meaning with the words.The Word Sense Disambiguation is the process of finding correct senses of a word, among other senses associated with the words.The introduction of semantics in word level to improve Word Senses Disambiguation has been considered in this thesis specifically to improve the accuracy of WSD and thus in turn to improve the IR performance.In this work, the glosses of the indexed words in WordNet are utilized as conceptual information, which acts as an additional semantics for WSD.This concept based WSD, has been used in semantic chaining to cluster documents, which is used for IR performance.