AN ENHANCED SEMANTIC INFORMATION RETRIEVAL SYSTEM USING ONTOLOGY-BASED ALGORITHM

Well Researched and Ready to use Ph.D Thesis, page numbers: 131, Department: Computer Science

ABSTRACT 

The web is a large information hive, where information is stored and shared. Over time, the amount of information on the web keeps increasing steadily and fast. This geometric growth has brought about difficulties in searching for the exact needed information from the web. Search engines have been the main tool used for searching information on the web. They collected, stored and pre-processed information on the web as indexes. Recently, there has been a great improvement with the development of algorithms used by the search engines. However, users still need to put in considerable efforts in order to access relevant information because the supporting technology does not make it simple enough to add ontology-based metadata to information, and the documents are only retrieved based on the keywords. Hence, this study presents an effective ontology-based system, which index documents according to concepts that best describe them, and a retrieval module that optimally utilizes the ontology tools to improve on both recall and precision. The specific objectives are to: (i) evaluate the existing Information Retrieval (IR) algorithms in order to discover their strengths and weaknesses; (ii) develop an enhanced ontology-based algorithm for IR System; (iii) investigate the effectiveness of query expansion when an upper-level ontology is combined with domain-specific ontology; and (iv) conduct performance evaluation of the developed algorithm with the existing algorithms based on Recall and Mean Average Precision. The method used involved the concept of Ontology Knowledge Bases (OKB) as document repository for fast retrieval. The indexing module implemented automatic semantic indexing using OKBs for both upper-level and domain-specific ontology. The Classic Vector-Space Model was adapted for retrieval module. Integration of Query Expansion as to unbound all constraints of original users’ query and a ranking algorithm were employed. The findings of the study revealed that: (i) taking full advantages of ontology in the development of IR System enhanced its performance, irrespective of the domain; (ii) ontologies bridged the gap between query terms and documents through semantic mechanisms as given in the equation below: where E(o,T) and P(o,T) are the number of classes of ontology o that have labels that match any of the search-texts exactly or partially, respectively; (iii) added Query Expansion was shown to be flexible since all the constraints are unbounded; and (iv) values of recall and precision were very good when compared with two other existing IRs (Lucene and Sphider). The value of recall and precision were between 0 and 1. In this study, it was concluded that the added value of semantic information retrieval over traditional keyword-based retrieval has helped to achieve better precision through query weight and better recall through semantic relations. The proposed system is robust, flexible and efficient. The results of the study can be useful in digital libraries, information filtering, media search and search engines.

ORDER the Complete Ph.D Thesis

Share!!