Bayesian graphic model based user preference prediction for future personalized service provisioning
Ying Wang, Peilong Li, Haiqing Tao, Rui Meng, Jiajun Liu · 2015
In the era of big data, gathering data becomes increasingly cheaper, and more and more different types of data is capable to be collected and stored. Accordingly the traditional user independent service provisioning is no longer satisfying. Further performance improvement can be achieved by making use of the collected personalized data. In this paper we pursue predicting user preference given the personalized data. A Bayesian Graphic Model is proposed accordingly. Because of the nature of proposed model, there is no closed form solution for optimization of model parameters. An iteratively expectation maximization (EM) algorithm is therefore employed for model training. Furthermore, a Monte Carlo method is also used to simplify the calculation in the expectation step. In order to demonstrate the effectiveness of our approach, a MovieLens [1] data set is used and the experimental results show that the performance of the proposed approach has a significant performance improvement comparing with the traditional method.