Designe And Realize The Personalized Recommended Search Engine

Along with the rapid development of modern e-commerce and network, each largecommerce site has provided more humanized and personalized service for users, andresearches on personalized recommended search engine have been widely applied. Atpresent, the personalized recommended engine in each large commerce site isrecommended the same or aiming at popular category, rather than designing a kind ofrecommended search engine to offer different information for different users. This kindof search engine is just like the commodity salesman knowing your character andinterest, who can recommend needed commodity information to you at the first time.In view of above problems, this thesis has designed a simple personalizedrecommended search engine system supporting short text including OnLine module andOffLine module. According to different registered users, the system can offeredpersonalized recommended information aiming at users’ interest. The system in thisthesis analyzes the following main contents:OnLine system module, including several parts of Http server, query parser, userprofile obtainer, Rank mark core mechanism and mark sequencing. A simple Http serveris designed to be a service container in the system. The personalized recommendedsearch engine system researched and designed in this thesis is a lightweight system,therefore, it needs a same simple and lightweight Heep network server to support. Withuser profile obtainer, we can obtain users’ basic information and interest. With queryparser, we can obtain a inverted list after the treatment of related records in the queryrecord. Rank mark core mechanism is also the core of system, having the mark processon all data record obtained from inverted list and user profile obtainer. According tomark sequencing, the advanced result set obtaining a certain amount is therecommended result set based on users’ interest and purchasing history, thus returning torecords for larger correlation with priority and users’ interest.OffLine data processing module, responsible for background data processing andsaving to storage system. The data needs processing in the background are mainly userdimension module and query dimension module. User information can be divided into basic information and behavior information. Basic information can be obtained withexperimental data, including the most fundamental feature for user. Behaviorinformation can be generally divided into two parts of interest and purchasing history,reflecting users’ interest. With the analysis on users’ information and query parser, theinvalid record should be filtered.