Research On Personalized Semantic Search Driven By Ontology Evolution

With the surge in the amount of information on the Internet and the rapid growth of the number of search engine users,the limitations of the traditional Web search engine gradually exposed? On the one hand,because the traditional search engine provides a "search oriented"rather than "user oriented" service,users often face the problem of "information overload" and"resource oriented" problem.On the other hand,due to the use of keyword based character matching rather than semantic based information search technology,the traditional search engine faces two deeper problems and challenges:"faithful expression" and "expression difference".It is in this background and demand,personalized search engine and semantic search engine should be potential and out,used to improve the traditional network search technology.Nowadays,personalized search engine and semantic search engine are the hot spots in the research of new generation search engine.The quality of the search result is crucial for the success of a search system and depends mainly on the domain ontology in semantic search engine.However,the domain ontology usually fail to be updated in time and still stay in the initial state that does not contain enough semantics to provide the powerful support for semantic search.As a result,the quality of the search result was often poor.The thesis mainly focuses on personalized semantic search and ontology evolution,and the specific content is as follows:(1)The research on personalized semantic search method combining personalization with semanticIn the light of the problem of topic drift in personalized search and the defect of'retrieval oriented' rather than 'user oriented' in semantic search,a new way of combining personalized search with semantic was presented,which fuses the personalized elements and the objective elements in semantic,realizing the "user oriented" search and effectively solving the theme drift phenomenon.Based on the nearest neighbor set method in collaborative filtering,a user group clustering algorithm is proposed,which can achieve the heuristic extension of the search keywords.The way of multi-dimensional personalized semantic search and algorithm of multidimensional re-ranking proposed in this research based on Web log mining,user clustering and concept similarity can effectively improve the accuracy of search results and the accuracy of search results ranking,to promote the satisfaction of user search.(2)The research on new method of ontology evolutionIn view of the defect of existing ontology based semantic search engine improving or correcting the ontology in the system only through the combination with other ontology instead of using domain knowledge to evolve Ontology,I proposed a real method of evolving ontology in system with new semantics directly from the network search that can be the ontology in any interest domain.The user's search results are formed knowledge element group by pretreatment of webpage analytics,de-duplication,content extraction,noise elimination,query ontology matching,etc,which are packaging into a corpus as the source of ontology evolution by corpus wrapper.Ontology evolution process includes two stages:ontology classification learning and ontology merging,forming directed acyclic classification graph by ontology classification learning and fusing new obtained semantics with the ontology in system by ontology merging.That can use the new knowledge in the field to continuously evolve Ontology to provide sufficient semantic support for the semantic search of high quality.(3)The research on the solution of ontology heterogeneityIn view of the problem of only solving ontology heterogeneity partly in ontology mapping,a multi-strategy mapping method based on structure mapping,syntax mapping and machine learning is proposed that can simultaneously deal with the inconsistency in the interpretation layer and the inconsistency on the concept layer,which fully solves the heterogeneous problem.(4)Realization of personalized semantic search engine prototype system and application researchBased on the foundation of above research,I designed and implemented a personalized semantic search engine system prototype driven by ontology evolution.I conducted a comprehensive test in the application environment with it,and verified the theoretical results by experiments.The theoretical results of personalized semantic search and ontology evolution in this research have certain theoretical and practical significance to explore and solve the intelligent,humanized,and knowledge search experience and service.More importantly,some of the techniques and methods proposed in this study have a wide range of applicability and reference value for the research of Web search.