A Stock Secommendation System Using with Distributed Graph Computation and Trust Model-Collaborative Filtering Algorithm
Haoyu Wang, Yi Sun, Xujie Li, Yonghong Xie, Yuxin Qi · 2018
Using a recommender systems is an effective way to solve the problem of information overload and help users to discover valuable information. This paper builds a dichotomy model of shareholder-stock relationship based on the distributed graph computing framework Spark GraphX, and using a certain financial theory, transforms the investment behavior of users (shareholders) into the ratings and trust of the invested stock. Then we calculate the shareholder similarity graph and the trust graph of shareholders through the parallel graph calculation, and use the improved collaborative filtering algorithm based on trust model to make a recommendation analysis of the stock of A-shares and SME stocks in the stock market. Finally, the comparison of common collaborative filtering algorithm in the experimental environment shows that the system has better algorithm scalability and accuracy.