The Research And Implementation Of Collaborative Filtering Recommendation System Based On Hadoop

With the rapid development of the Internet,Some of the technological products that change people's lives are derived.One of the most representative products is recommendation system,which makes people no longer find the information they need from the mass data through the search engine in the past,but select the items of information provided by the recommender system.The change of this situation is due to the results of the research on the recommendation algorithm.Collaborative filtering recommendation algorithm is the most widely used algorithm in the field of recommendation.In the process of collaborative filtering algorithm applied to the actual scene,data cannot be recommended for new users and not considering the context factors make such problems as lack of personalized recommendation results exposed gradually due to dependence on user's behaviors.Based on the collaborative filtering recommendation algorithm,this paper proposes a hybrid recommendation algorithm based on user interest change and user characteristics,which is called ICUAH algorithm.The main work is as follows:Study the similarity measure method and the typical recommendation algorithm commonly used in the recommendation system deeply.On the basis of this,realize the traditional collaborative filtering recommendation algorithm and compare with the two algorithms through the actual scene,which lays the foundation for the improvement of the algorithm.This paper combine with the classical TF-IDF algorithm and the alternating least squares algorithm to present a brief solution to the problem of data sparsity and item cold start in the traditional collaborative filtering recommendation algorithm.The traditional collaborative filtering recommendation algorithm is combined with the time context factor and the user characteristic attribute.On the one hand,the similarity measurement and the users' preference are improved.The solution to the penalty of the popular item is put forward on the other hand.A hybrid recommendation algorithm based on user interest change and user characteristics is proposed.The ICUAH algorithm is implemented by the Java language and is evaluated experimentally using the Movie Lens data set.The results of the experiment show that the ICUAH algorithm can improve the accuracy and coverage rate.The ICUAH algorithm is parallelized by the Map Reduce which is a distributed computing framework.Based on the ICUAH algorithm,a small book recommendation system is designed and implemented using the Spring Boot framework and the Hadoop distributed framework.