Data Mining and Knowledge Engineering

Abstract

Relevance storing and searching are key to web search in determining how results are retrieved and ordered. As keyword-based search does not guarantee relevance in meanings,semantic search has attracted enormous and growing interest to improve the accuracy of relevance ranking. Recently heterogeneous semantic information such as thesauruses, semantic markups and social annotations has been adopted in search respectively for this purpose. However, although to integrate more semantics would logically generate better search results in respect of semantic relevance, such integrated semantic search mechanism is still in absence and to be researched. This paper proposes multi-agent based semantic search approach to integrate both files and heterogeneous semantics. Such integration is achieved through semantic relationship between files. This research helps us to reduce the size of the file while storing and the files are stored based on a certain relationship. These files are compressed. Similar types of files are stored in specific drive of the hard ware. With respect to the great volumes of distributed and dynamic web information, this multi-agent based approach not only guarantees efficiency and reliability of search, but also enables automatic and effective cooperation’s for semantic integration. Experiments show that the proposed approach can effectively integrate both keywords and heterogeneous semantics for web search. Search engines are useful tools in looking for information from the Internet. However,due to the difficulties of specifying appropriate queries and the problems of keyword-based similarity ranking presently encountered by search engines, general users are still not satisfied with the results retrieved. To remedy the above difficulties and problems, in this paper we present a multi-agent framework in which an interactive approach is proposed to iteratively collect a user's feedback from the pages he has identified. By analyzing the pages gathered, the system can then gradually formulate queries to efficiently describe the content a user is looking for. In our framework, the evolution strategies are employed to evolve critical feature words for concept modeling in query formulation. The experimental results show that the framework developed is efficient and useful to enhance the quality of web search, and the concept based semantic search can thus be achieved. This system,agent based web browser, resides on users' computers providing with effective advice to help them locate the relevant information required from their browsing experience to view www document accessing the Internet taking advantages of text formatting,hypertext links, images, sounds, motion, and other features. Agent based web browser, is compatible with modern web pages and effectiveness and efficiency technologies of the process, active security features and the fastest response times, thousands of free ways to personalize user's online experience, superior speed and performance and autonomous process, running in the background of the computer while the user is the one with absolute control over the browsing path. The new generation of browser will be smarter by working online and offline to facilitate the connections of using ideal AI agents to be communicative, capable and autonomous able to understand user's goals, preferences and constraints. And it must be able to act without the user being in control the whole time.

Keywords

Full Text

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.

Keywords

Full Text