A Man-machine Interactive Search Engine System Based on the Relevance Feedback
|School||Dalian University of Technology|
|Course||Applied Computer Technology|
|Keywords||Search Engine Vector Space Model Relevant Degree Feedback|
Rapid development and popularizing of Internet causes online information to increase fast extensively. How to obtain valuable information on huge Internet becomes the question that users pay close attention to day by day. The technology of search engine’s appearance has found necessary information and provided convenience fast for users. The search engine is a kind of search tool used for helping the Internet user to inquire about information. For the purpose of information navigation, it collects and finds information in Internet with certain tactics, understand, draw, organize and deal with information, and offer service of searching for users. As the technology of the search engine is being developed forward constantly, the search engine becomes the user main tool used for searching online information.But the search engine has much limitation, e.g. the rate of query overall is low; the rate of query accurately is low; the result users search for and ask for difference bigger; the grammar of every search engine is not unified and users are difficult to master them; lacking the special topic search engine specially facing a certain discipline; the intellectual degree of the search engine is very low. So the search engine keeps making great efforts to improve. The technology of the search engine is becoming the target industrial circle of the computer and academia are falling over each other to study, develop. The purpose of research is that in case of not improving the degree of difficulty and efficiency users search, the search engine tries its best to retrieve documents including more information users want. The paper is just for this research purpose. The paper design and develop a interactive search engine based on vector space model to realize users’ individualized inquiry and improve the accuracy of inquiry. It uses the search results users think as " ideal" and " unsatisfactory " to be feedback information, draws model based on users’ " ideal " and " unsatisfactory" vector information, get text vector of the characteristic users think " ideal " and " unsatisfactory", get pages close to users’ purpose according to the dependence between them and users’ purpose, and refer them to users. It improves the rate of accuracy of the search engine system, prevents users from assigning the order in the form of inputting the keyword, and make inquiry more intelligent and humanizing.