融合主题模型和协同过滤的多样化移动应用推荐

随着移动应用的急速增长，手机助手等移动应用获取平台也面临着信息过载的问题.面对大量的移动应用，用户很难找到想到的或适合的应用，而另一方面长尾应用淹没在资源池中不被人所知.已有推荐方法多注重推荐准确率，忽视多样性，推荐结果中多是下载量高的应用，使得推荐系统的数据积累越来越偏向于热门应用，导致长期的推荐效果越来越差.针对此问题，本文首先改进了两个推荐方法，提出了将用户的主题模型和应用的主题模型与MF相结合的LDA_MF模型，以及将应用的标签信息和用户行为数据同时加以考虑的LDA_CF算法.为了结合不同算法的优点，在保证推荐准确率的条件下提升推荐结果的多样性，我们提出了融合LDA_MF、LDA_CF以及经典的基于物品的协同过滤模型的混合推荐算法.文章使用真实的大数据评测所提推荐算法，结果显示所提推荐方法能够得到推荐多样性更好且准确率高的结果.

Diversified Mobile App Recommendation Combining Topic Model and Collaborative Filtering

With the rapid growth of mobile applications,users of mobile app platforms are facing problem of information overload.Facing a large number of apps,users feel hard to find approporiate apps and a large number of long tail appsaresubmerged in the resource pool,unknown to most users.Meanwhile,existing recommendation methods usually pay more attention to accuracy than diversity,with most recommended items being popular ones.As a result,the overall exposure rate of mobile appsis low and behavior data accumulated by the system is gradually towards popular apps,which lead to a poor recommendationperformance in the long run.To solve this problem,we first propose two recommendations methodsnamed LDA_MF and LDA_CF respectively to improve existing methods,where LDA_MF combinesuser topic model and app topic model with matrix factorization model MF,and LDA_CF takesboth taginformation of apps and user behavior data into consideration.In order to take advantages of different algorithms and increasethe diversity of recommendation results without sacrificingaccuracy,we proposed a hybrid recommendation algorithm to combine LDA_MF,LDA_CF and item-based collaborative filtering models.We use a large real data set to evaluatethe proposed methods,and the results show that our algorithm can get betterdiversityand good recommendation accuracy.