基于融合标签与蚁群的协同过滤微博推荐算法

Micro-blog Recommendation Algorithm Based on Collaborative Filtering by Combining Tag and Ant Colony

针对协同过滤算法中存在数据稀疏的问题,提出一种基于融合用户标签和蚁群的协同过滤微博推荐算法。将表示用户兴趣的标签引入推荐模型中,利用标签和用户以及标签和微博的关联度,建立用户对微博的兴趣度模型。另外结合蚁群聚类和协同过滤为目标用户进行用户聚类,计算出对目标用户的待推荐微博集。最后利用用户对微博的兴趣度模型从待推荐微博集中选出Top-N为目标用户进行推荐。实验引入标签和蚁群算法的有效性,将测试结果与传统协同过滤推荐算法和纯基于标签的微博推荐算法进行比较,该算法不仅改善了协同过滤算法中数据稀疏和冷启动的问题,而且推荐准确度有明显提高。

Aiming at the problems of sparse and cold start of data in micro-blog recommendation,this paper proposes a fusion tag and ant colony collaborative filtering micro-blog recommendation algorithm.The tag will be interested in the recommended model to establish a user interest rate model of micro-blog by using the relevance of tag-user and tag-blog.In addition,ant colony clustering and collaborative filtering for the target user clustering are used to calculate the recommended micro-blog set for the target users.Finally,the user＇s interest rate model of micro-blog from the recommended micro-blog is focused on Top-N microblog recommended for the target users.The results show that compared with the traditional cooperative filtering recommendation algorithm and the purely based micro-blog recommendation algorithm,the algorithm not only improves the data sparse and cold start in the cooperative filtering algorithm of the problem,but also the recommended accuracy is improved significantly.