Study On Search Results Clustering Based On Formal Concept Analysis

At present, the existing search engines adopt various kinds of methods to improve the precision of the search result but the search result inquired about and asked by users still include the irrelevant file. Though sort through relevant degree, the relevant and irrelevant file that yet mix each other have brought to user burden. In order to help Web users to filter the file that is really interested in from a large number of results returned from search engine, search engine results are carried on clustering according to theme, so as to cut down result quantity that have been looked around by user and improve users' search efficiency. On the foundation of research and analysis to the process of clustering the search engine result, the Chinese key phrase extract, the onstage search, clustering algorithm is simplified and improved, search engine system which suited the user is produced. First of all, Meta Search Engine is realized, it can make use of a lot of search engine existing to obtain the relevant information on Internet, thus the resource coverage rate that users search greatly improved. Then word segment tool is utilized to carry on Chinese word segment and pretreatment to result returned from Meta Search Engine, improved C-Value/NC-Value algorithm is utilized to extract the nested key phrase from what has been returned in the passage. Finally, incremental algorithm of producing concept lattice is used to carry on concept clustering to the passage of search results, and produced the theme of each cluster result from it. Among them, the noun + extract way that suits Chinese is proposed, in order to reduce the loss factor, stack structure processing extract noun is proposed. when noun is confirmed for term quantifying term performance is executed by candidate term appearance frequency, times candidate term is inlaid by others, the number of term that inlays current candidate term, candidate term length, context information, key phrase is obtained with that.In the kind of process of the clustering result, the concept lattice construction method is produced, according to the object and the attribute mapping relations in the form background establish dictionary tree, finally the concept cover graph is obtained. The whole clustering of the search result is finished.