基于用户的协同过滤推荐算法的改进

The Improving of Collaborative Filtering Recommendation Algorithm Based on User

基于用户的协同过滤算法是根据用户对项目采取的行为来学习用户的兴趣和爱好,利用学习的结果给用户推荐其感兴趣的商品。传统的基于用户的协同过滤推荐算法通过用户购买、浏览、收藏等操作来计算用户之间的相似度,然后对相似度按照从高到低的顺序排序,找出相似度最高的N个用户作为邻居用户,把邻居用户所购买并且该用户尚未购买的商品推荐给用户,该算法只考虑了用户和商品的关系,忽略了用户与用户之间可能存在的潜在关系。针对以上问题,本文提出一种改进的用户协同过滤算法,首先根据用户的购买的商品求出用户之间的相似度,接着根据用户的地理位置计算用户之间的相似度,然后把二者加权求和,实验结果表明,改进后的推荐算法准确率得到了提高。

collaborative filtering algorithm based on user-item matrix , get people’s interest and hobby according to people who take action to the item ,and then recommend some items to people which they are interesting in . The traditional user-based collaborative filtering recommendation algorithm calculates similarity between users through operations such as user purchasing , browsing and collecting ,then sorts the similarity from high to low, finds the N users with the highest similarity as neighbors , and recommends the products purchased by the neighbors and the users who have not yet purchased . However , the algorithm only considers the relationship between the user and the product , ignoring the potential relationship between the two users . Aiming at the above problems , this paper proposes an improved user collaborative filtering algorithm. Firstly, the user's purchased goods are used to find the similarity between users. Then the similarity between users is calculated according to the user's geographical location , and then the two are weighted . Summation , experimental results show that the improved accuracy of the proposed algorithm has been improved .