Implementation Of Distributed Recommendation Algorithm Based On Improved Restricted Boltzmann Machine

With the rapid development of information technology, the Internet is becoming more and more popular, and the information on the Internet is growing exponentially, the problem we face is not the poor of information, but the fact that information overload. From a large number of network information, it is difficult for users to find the valuable information quickly and accurately. Facing the increasingly prominent contradiction between the information overload and the personalized demand, recommender system has become an effective means to solve this problem. Collaborative filtering algorithm is the most widely used recommendation technology, but with the expansion of network scale, the collaborative filtering technology is facing many challenges, its efficiency is still not up to expectations. In the face of massive data, the traditional recommendation system has been unable to solve the problem of information overload effectively.At present, collaborative filtering algorithm use Restricted Boltzmann Machine based on the the same interests and preferences of users to make recommendations, but with the rapid growth of information and data, the efficiency of the algorithm is lower an lower. In this view, this paper uses the method of adding impulse to increase speed in Recommendation system, and the application of MapReduce technology to achieve parallelism of collaborative filtering algorithm based on Restricted Boltzmann Machine so as to further improve the efficiency of the algorithm.This paper mainly gives the methods to improve the efficiency of the recommendation system, and uses MapReduce parallel computing model to improve the efficiency of recommendation. The following is the main content:firstly, this paper describes the current popular recommendation algorithm, and then the model of algorithm, also the advantages and disadvantages are introduced. Then introduce the Hadoop platform, including the HDFS and MapReduce; then, introduces the Restricted Boltzmann Machine algorithm, proposed the method to improve the efficiency of the algorithm, which is adding the impulse, and on this basis, the combination of MapReduce to collaborative filtering system for parallel design in order to further improve the efficiency of recommendation algorithm. Finally, this paper sets up an experiment on distributed platform to verify the calculation efficiency in MovieLens data, compared with the traditional collaborative filtering algorithm,experiments show that a collaborative filtering recommendation with improved Restricted Boltzmann Machine based on distributed platform can improve the efficiency greatly.