AN IMPROVED REFERENCE SEARCH MODEL AND RELATIVITY TECHNOLOGIES

The web based reference search is getting more and more popular. The general reference search provide some advanced search methods, such as topic based, author based, organization based, keyword based, abstract based, publication based, quotation based and fund based search methods, and it also provides the full text based search method. These search methods are all Boolean methods, and all the articles where the query string exists are regarded as the search results. This leads to the poor search quality. The relativity based sorting method can sort the search results by the relativity between the query string and the article, and then the quality of reference search can be improved. But the reference relativity can not be given accurately by the current calculating methods; thereby the quality of reference search is affected. First proposed in this paper is a reference structure model in which the concept of virtual nodes is brought forward and the scope of the reference search is extended greatly. Secondly, based on the reference structure model, the citation relations among the references are mined to calculate the importance of the references in the reference repository. Finally, The precision and the quality of the reference search are improved greatly by combining the foregoing two methods.