融合项目因素的用户部分特征协同过滤算法

User Feature Collaborative Filtering Algorithm Integrating Item Factors

为解决基于用户部分特征协同过滤算法中出现的项目分类不准确问题,本文提出了一种融合了近似项目的用户部分特征协同过滤算法(AICF)。该算法通过融合项目标签数据和用户评分数据来计算项目综合相似度,以获取待推荐项目的近似项目,然后在近似项目中计算待推荐用户最近邻,最后根据近邻用户对指定用户推荐。实验表明,通过结合项目标签数据改进近邻项目的选择,可以提高最终协同过滤算法的推荐效果。

In order to solve the problem of inaccuracy of project classification based on user partial feature collaborative filtering algorithm, a kind of user partial feature collaborative filtering algorithm is proposed, which combines the approximate item (AICF). The algorithm combines project label data and user score data to calculate the comprehensive similarity of the project to obtain the approximate items of the recommended items. Then, the nearest neighbor is calculated in the approximate project. Finally, the recommended user is recommended by the nearest neighbor to the designated user. Experiments show that by combining the project label data and improving the selection of neighbor items, the recommendation effect of the final collaborative filtering algorithm can be improved.