Research On The Parallel Hypernetworks Model And Its Application In Movie Score

Score prediction problem is the core problem of the recommendation system. In order to improve the prediction accuracy of recommendation system, a recommendation algorithm based on the Hypernetworks model is proposed. The algorithm uses the Hypernetworks to model the users' rating matrix and fully excavates the users' interest factors. The results of experiments, which are carried out to process the real world data sets, indicate that the recommendation algorithm based on the Hypernetworks model has a significant improvement in the accuracy of prediction.However, due to the limitation of its own model, Hypernetworks model has the disadvantage of high time complexity, especially when the high dimensional data are processed. The problem is difficult to be solved in the serial computational mode. The paper presents a concurrent model of Hypernetworks to make full use of the multi core resources of CPU. Furthermore, this paper presents a parallel computational model of Hypernetworks under the cloud-computing platform Hadoop.The single machine algorithm is transformed to be executed on multiple machines. The experimental result shows that parallel Hypernetworks model based on cloud-computing platform Hadoop has better scalability and a better time efficiency, which turns out to be the main advantage of the parallel Hypernetworks model.The main work and achievements are as follows:1. This paper proposes a recommendation system based on the model of Hypernetworks and uses the model to deal with the movie score prediction problems. The related experimental results show that this model has a great advantage in prediction accuracy.2. Multi-thread technology is used for concurrent programming of the Hypernetworks model and the computation time is decreased.3. In order to solve the problem of high computational overhead caused by the Hypernetworks model when dealing with large data sets, this paper presents a parallel model of Hypernetworks, which transplants the traditional Hypernetworks model onto the cloud platform Hadoop for distributed computation. The related experimental results show that the parallel model has a great advantage in time efficiency and classification accuracy.