Research on the Modeling and Related Algorithms of Label-Weight Rating Based Recommendation System

Recommendation System has been frequently applied into various e-commerce websites and social networking sites. With improving users' satisfaction, recommendation system has also brought huge commercial interests. However, as the original data is incomplete and some recommendation algorithms have their own special way of processing data, current recommendation system sometimes cannot work very well.For example, some recommendation systems are bothered with cold-start problem, difficult for complex interest recommendation problem, poor interpretability and so on. Consequently, in the paper, we propose a recommendation system modeling based on label-weight rating. In this system, first we will get the most accurate evaluation and demanding information of users in a more concise way-label-weight rating method. Then we will generate recommendations using improved existing recommendation algorithm. Finally, we will show the recommendations to the users in the form of label-weight rating and make reasonable explanation to users. In the extended experiments we design a series of movie recommendations experiments to prove the effectiveness and feasibility of the modeling.