Research On Movies Personalized Recommendation System Based On Users Access Data

For nearly 10 years,as the Internet gradually spread in society,the information explosion phenomenon has become increasingly apparent.Users from all walks of life keep providing Internet with information,making the Internet an all-encompassing information gathering body.Meanwhile the Internet user is difficult to quickly find something that meets their own needs and interest in the vast ocean of information.People will obtain same results when retrieving information using a search engine with the same keyword,but user's demand for information is diversified and personalized.Therefore,the traditional homogeneity of search engines as a representative of an information retrieval system has been unable to meet the individual needs of users,so personalized recommendation system comes into being.Personalized recommendation system extracts records of interest related by mining user's historical behavior data,calculates user's interest through algorithm of cetain rules,and recommends user with information meeting their interest,solving the conflict between large amount information and difficulty of chosing.Recommendation system keeps tracking user's behavior history,updating and iterativing user interest,so that information recommended to users always fit user's interest,which makes user more easily access the information they are interested in.The ultimate goal of recommendation system is to achieve full customization for the user's information push.This article describes how to build a complete movie knowledge map,which can structured describe user's behavior.According to user's viewing behavior characteristics and properties of the film itself,the movie is divided into independent films and series of films,which helps to build Movies knowledge map more fine-grained.This article proposes an algorithm to series of excavations.The basic data used for research is user's access request,which makes it possible to extract user's interest without user's participation,avoiding the incomprehension from user's subjective choice.Extracting the data associated with the movie through the analysis and processing of the user's original access request,user's behavior can be described with machine language,so as to achieve extracting user's interest.Building a user model vector based on TF-IDF algorithm combined with time decay,it can be possible to calculate user's total interest degree in this movie by calculating the degree of user interest in each dimension elements of movie.Finally,experimental analysis demonstrated high rates of recall and precision referred to herein schemes.