Research On Campus Educational Resources' Personalized Recommendation Based On Collaborative Filtering

With the expansion of the Internet information,WWW has become a service website that contains variety of information resources,it provides the user with a very valuable information.But with the increasing amount of user access network,it has become increasingly hard for users to find the information they want in the network which is like a ocean,how to make our customers convenient,rapid,accurate to obtain the information has become the major problems that solved by the websites, personalized recommendation technology development and utilization is the important way to solve this problem.As the personalized recommendation systems technology promotion becoming increasingly popular,and gradually become the research hotspots,because the usage of active recommended technique,active recommending resources,satisfy customer demand,many users do not willing to spend too much time in a website.Personalized recommendation,as a kind of new intelligent information service mode,can provide information and service accurately to users according to their requirements,or clear by a user's personality,habits and preferences,thus it solved the problems that brought by information overload and user lost.This paper do some research based on collaborative filtering technology,puts forward the specific method to construct personalized educational resources recommendation website.The paper analyzes the causes of website construction and feasibility analysis,according to the parts of collaborative filtering technology such as data input,neighbor formation,resulting in recommend,corresponding information resources and students interested in model,For evaluation matrix,score predicts,Do different kinds of reommendation of different category resources in the process of recommendation.In the process of recommended according to different category, improve the traditional slope one algorithm,increases the user and projects in order to recuce the similarity calculation and increase the prediction accuracy.To different categories,resources are recommended and the accuracy will be improved.According to the different categories of different resources recommended, evaluation method is improved,abandoned by the number of resources for the traditional method,the unit by keyword is considered.Combined with the user access recordset and recommended resources,the calculation precision is improved.In the end,the essay do some summary and analysis the content which need to improve and in-depth study.