Research Of Ranking Algorithm Of Personalized Search Engine

In recent years,the search engine has become one of the popular application of Internet,and it is also an important way for people to obtain information,resources.But general search engine uses traditional ranking algorithm and cannot accurately get users' search intention,and the search results is "one size fits all" when different users query one word,rather than personalized results for users,so that users have to spent more time searching repeatedly in it.Thus,in order to improve the ranking result of search engine,user-oriented personalized ranking algorithm is becoming one of the hottest research directions of search engine.A new personalized ranking algorithm based on interest forgetting is designed,and the main work in researching is as follows:1.Researching on the current situation of development of ranking algorithm and personalization technology in search engine at home and abroad,and gaining insight into the structure of search engine,ranking algorithm and personalization technology,and analyzing that advantages and limitations of traditional ranking algorithms and existing personalized ranking algorithms.2.Page analysis and feature extraction as updating User Interest Model or ranking pages in existed personalized ranking algorithm,which will have a bad influence on retrieval efficiency and feature extraction.So Page Feature Model was designed to store the features of pages of the index database in advance.3.In view of the degree of search interest attenuating over time is not taken into existing personalized ranking algorithm,and have a bad influence on predicting user interest,a search interest forgetting algorithm,which can simulate the forgetting process of user interest,was proposed by researching on forgetting curve of H.Ebbinghaus and changing law of search interest.4.Based on the elaborate analysis of creating and updating process of existing user interest model,a user interest model with search interest forgetting algorithm was designed,which can fix weight of interest in updating and improve existing user interest model.5.Considering the user features in searching and attenuation feature of search interest,a re-ranking algorithm was designed to re-rank the result pages generated by traditional search engine,and a personalized search engine could be constructed to provide a personalized search results for user.Search engine system was built with Lucene development kit,and some experiments was done with traditional ranking algorithm,traditional personalized ranking algorithm and this algorithm.Experimental results show that this algorithm can provide the search results more in line with the intention of users compared to traditional ranking algorithm,and improve users'satisfaction.And compared to traditional personalized ranking algorithm,this algorithm can predict users' search intention more accurately,and improve precision of search engines.