The Research Of Personalized Search Engine Base On Fuzzy Rough Sets

With the Internet being used more and more widely in human society, the amount of information in the Internet is increasing rapidly. The amount is so large that it makes most people difficult to get the information that they need immediately, thus, they have to use engine searches to help them to search out these information. However, the existing search engines for different users always feed back the same result, without considering individual differences between users, which can cause that a part of users have worse experience than other users when searching ambiguous words. According to the survey, China’s users are usually only interested in the first page, which requires search engines to upgrade the search results that conform to users’interest, on the other hand, most users’interests and habits have been considered in the result from search engines, and users might have temporary interest in other fields, so it is not appropriate to negate the existing ranking result of search engines completely, because users’interests may be overconformed and they may only see the results of one field.This paper proposes a personalized search engine reorder algorithm based on fuzzy rough sets. This algorithm reorders the original data from Baidu to achieve the goal of personalized search service. Firstly, some relevant knowledge of personalized search engine is introduced in this paper; secondly, a two-level user interest profile is proposed, in which the first level is considered as the upper approximation set of fuzzy rough sets, while the second level is considered as the lower approximation set; thirdly, a reordering algorithm based on the similarity algorithm of fuzzy rough sets is proposed. It calculates the similarity of the results from Baidu and user interest profile, and then combines the original ranking result from Baidu to present a new ranking result; finally, experimental results of this personalized search engine algorithm are cited and analyzed. The results show that the algorithm can meet the individual needs of users.