一种基于混合相似度的用户多兴趣推荐算法

User Multi-Faced Interests Recommendation Algorithm Based on Hybrid Similarity

针对传统协同过滤推荐数据稀疏会影响推荐质量,以及项目最近邻居集的计算忽略用户多兴趣及提高推荐的准确度问题,该文采用混合模型改进了相似性度量计算,综合Pearson相关系数与修正余弦相似性,提出了一种基于混合相似度的用户多兴趣推荐算法.实验表明:该推荐方法的相似度计算更高效,不仅提高推荐准确率,而且使用户有更好的推荐体验.

The traditional collaborative filtering recommendation＇s sparse data will affect the quality, and it fails to take into account the user multi-faced interests to determine the projects nearest neighbor set. Coupling with the traditional similarity measure method without considering user＇s behavior, leads to lower quality of the recommendation. In order to improve the recommendation accuracy, the hybrid model, improved similarity measure calculated by Pearson correlation linear combination of adjusted cosine correlation has been used, and then an user multi-faced in- terests recommendation algorithm of hybrid similarity computing is proposed in the paper. The experimental results show that the similarity calculation of recommend dation method is more efficient, improve the accuracy of recommendation, and make the better recommendation of user experience.