Application Of Ontology In Information Retrieval And Its Evolution Mechanism Research

In recent years,with the development of information technology,the quantity of web resource increase dramatically.however, amount of overlaps among web pages emerge because of connecting to each other;it is difficulty to create Index resulting from the semi-structured or unstructured informations; Especially, in the age of Web 2.0, everyone are allowed to publish information on the Internet, as a result of the lack of constraint,the quality of information can not be guaranteed,content and format hardly reach the standard.when the Internet provide convenience benefits for human,the trap in the retrieval also arise.Although traditional information retrieval system develop continually,it become more difficult to meet the need of users in the complicated Internet enviroment. I found that there remains weakness in the information retrieval systems:the quantity of web pages that search engine returns is so much that users have no patience to visit page by page, most of users only browse the first three pages; Related Searches of search engine and query expansion of database are frequently unutilized.I also found by my experiments that the effect of the expansion is by no means satisfactory, the vocabulary that retrieval systems recommend have no semantic relation with users'keywords; Generally,user don not know their specific retrieval goal,even if they know,it is difficult for them to express precisely their needs in the round. Polysemy and Synonymy are universal in human language,it is impossible for users to master all polysemy and synonymy.But existent retrieval systems do not provide good query expansion and recommendation.As an important bridge between network information resources and users,retrieval systems have to consummate unceasingly so that recall and precision increase,which not only save users' time and energy,but also increase utilization rate of information resource. Throughout the process of information retrieval,the user's keywords are crucial importance,because it decide retrieval result.However, ordinary users'retrieval ability are limited,so intellectualized retrieval systems should realize query expansion and specification,and provide semantic recommendation for users.The appearance of ontology provide the opportunity to solve above problems,which is shared, conceptualized,explicit,normalized,formal specifications,are aimed at acquire all knowledge in domain,offer common understanding about domain knowledge and define clearly concepts and their relations in order to implement reasoning of domain knowledge.Ontology have been applied widely in the field of information retrieval and make some progress in the aspects of query expansion,information extraction, automatic classification, semantics formal representation and semantic reasoning have been a hot topic in the field of information retrieval. Famous ontology,Wordnet,have become one of the most popular ontology resource in query expansion,it can make retrieval results more comprehensive and precise.To resolve retrieval difficulty resulting from characteristics of network information, the limitations of users'retrieval behavior and weakness of retrieval system,on the basis of reviewing related research at home and abroad,this paper develop a new information retrieval system framework,which consist of user interface module,ontology reasoning module,ontology management module,Bayes Network computation module,server and database module.This system framework utilize strong knowledge base and reasoning ability of ontology to realize query expansion and normalization,use ontology reasoning technology to find out all neighbour nodes of original keyword in ontology and accomplish the transformation from keywords to shared concept in domain. Ultimately the query expandion are achieved.At the same time,this paper introduce Bayes Network to compute the semantic distance between original keywords and their neighbour nodes,and then submit this expansion set with ranking factors to matching module. Finally, the systems provide the retrieval results which are sorted by the sequence of nodes and the correlated degree between nodes and documents,and meanwhile concept set with ranking factors will be feed back to users as a recommendation.Traditional ontology construction and maintenance are implemented under the guidance of domain expert or knowledge engineer,which lead to high cost of ontology construction and maintenance and slow update speed.What's more,user totally separate from the progress of ontology evolution. So it is difficult to achieve wide participation and knowledge shared.Therefore,this paper advocate Web 2.0 idea, combine with the community collaborative ability of Folksonomy,make use of Wiki technology,and then this paper suggest to improve ontology under the dynamic and open environment which user can participate freely.Every participant can organize knowledge according to their actual need,creat link,edit class of ontology and talk about the meaning of concept with other users in the same community.The collaborative ontology evolution can optimize obviously ontology maintaining and promote knowledge sharing and flowing.