Agent Based Semantics Using Relationships

Abstract

The semantic web initiative of the world-wide web consortium (w3c) has been active for the last few years and has attracted interest and skepticism in equal measure. Semantic search generally deals with the searching of the keyword along with the concept related to that semantic word. Perhaps the most widely developed space at the moment within the semantic web is in information management, i.e. the organization and discovery of information. This is the primary motivation behind the semantic web’s development, but people are taking a variety of approaches to developing tools to extend the current web into a true semantic web. One of the methods of implementing semantic search is by means of agents like indexing agent, query agent and ontology agent. Index agent downloads files from web and extract textual content, query agent process all queries, ontology agent extracts the descriptive concepts and calculates relation between those concepts. The main problem with these agents is the pages retrieved after searching are not relevant. The other method of implementing semantic search is by combining semantic search and ontology learning. The need for domain ontology for information retrieval to improve answer queries, However ontology in information retrieval systems requires regular updates and relationship between new concepts. Semantic search and ontology collectively improves the value of indexing documents and enriching ontology. It is done by multi layer like domain ontology,topic ontology, and natural language ontology. The main constraint here is this system is based on little domain ontology that cannot be expanded. Semantic search is also implemented by improving results with light weight semantic search. Here search queries and light weight semantics are combined together to improve search results. It is done by basic and concept scoring computation. Here indexing is done allocating resources and results are obtained by means of search engines. The main problems in this research are the inverted resources and metadata are not allocated properly for indexing. The problems with the previous implementations are time complexity, accuracy in retrieving pages, domain ontology,and resource allocation for indexing. The semantic pages are searched in the query repository based upon the algorithm for query retrieval. Then an algorithm for ontology based search is implemented. The ontology is obtained based on multi level ontology structure. Then a retrieval algorithm is designed to implement the semantic search.