基于用户兴趣模型的个性推荐算法

Personalized recommendation algorithm based on user interest model

协同过滤推荐技术和基于商品属性的推荐技术是比较流行的个性化推荐方法,但是前者存在数据稀少和新对象问题,后者也存在无法挖掘用户潜在兴趣的问题。本文采用基于区域用户的相邻用户进行数据评分的矩阵填充,并采用商品之间的关联规则应用和解释来向用户推荐产品。测试表明,本方法解决了新商品的问题,并且在推荐的准确度、新颖性和覆盖度上有了较好的效果。

Collaborative filtering recommendation and recommendation based on product attributes are popular personalized recommendation methods. But the former can＇t handle the issues about sparse data and new objects,the latter is not capable of mining the potential interests of users. This paper uses the matrix of data grading based on the adjacent users of the regional users,and recommends the products to the users by applying and explaining the association rules between the commodities. Testing shows that this method solves the problem of new products and has good results in terms of accuracy,novelty and coverage.