一种多层混合的推荐模型研究

omprehensively analyzing the neighborhood model and the lingo model, we summarize the characteristics of existing technologies as well as the advantages and disadvantages of them. A hybrid Top-N model is established in order to effectively use the user feedback information and generally consider the potential relationship between users and items, by fusing the neighborhood model and lingo. With comprehensive consideration on the user feedback information, characteristic information, and potential information, we propose a novel multi-tier hybrid Top-N model based on SVD++, called SHT (SVD++-based Hybrid Top-N Recommender Model). It is able to recommend the most accurate and effective items to user by their preferences and habits. The experimental results show that each layer of the model can improve the accuracy of recommendation results in different extents.