An Improved and More Accurate Hybrid Recommendation Algorithm

推荐系统可以过滤一些无用信息，可以预测用户是否喜欢给定的资源。基于内容的推荐和协同过滤推荐算法是目前主要的个性化推荐方法。但是随着用户项目的不断增加，用户-项目评分矩阵存在着稀疏性、冷启动等问题。针对此问题，我们提出了一个独特的层叠混合推荐方法，使用评级数据，人口统计数据和特征数据来计算项目之间的相似度。实验表明我们的方法优于传统的推荐系统算法。 Recommender system can filter some useless information and can predict whether the users love given resources. Content-based recommendation and collaborative filtering recommendation algorithm is the main personalized recommendation method. However, with the continuous increase of user projects, there are sparse, cold start and other issues in the user-project scoring matrix. In response to this problem, we propose a unique cascade hybrid recommendation method that uses rating data, demographic data, and feature data to calculate the similarity between projects. Experiments show that our method is superior to the traditional recommendation system algorithm.