| 摘 要: 为了解决传统的协同过滤推荐算法计算用户之间相似性度量时,忽略用户与物品之间的相似关系导致推荐性能下降的问题,设计了一种结合遗忘机制与用户相似度的推荐算法。该算法基于用户-用户和物品-物品余弦相似度值和关系二元性,同时引入遗忘机制,根据用户对物品的评分以及记忆留存率进行偏好权重计算,再通过仔细合并相似度的值提高系统的覆盖率和点击率。通过在数据集MovieLens上与其他链接预测算法进行对比实验,结果证明该算法的命中率相较于其他算法提高了约7%,覆盖率略高于现有算法。 |
| 关键词: 推荐算法;余弦相似度;兴趣漂移;链接预测 |
|
中图分类号: TP312
文献标识码: A
|
|
| An Intelligent Recommendation Algorithm Based on Forgetting Mechanism and Cosine Similarity |
|
XU Xin, GUO Jiahe, QIAO Yu, SHU Wanneng
|
(School of Computer Science, South-Central Minzu University, Wuhan 430074, China)
2020120367@scuec.edu.cn; 2020120335@scuec.edu.cn; 2020120344@scuec.edu.cn; shuwanneng@whu.edu.cn
|
| Abstract: Traditional collaborative filtering recommendation algorithm tends to ignore the similarity between users and items when calculating the similarity measure between users, which leads to the decline of recommendation performance. In order to solve this problem, this paper proposes to design a recommendation algorithm combining forgetting mechanism and user similarity. The algorithm is based on the user-user and item-item cosine similarity values and relationship duality. At the same time, the forgetting mechanism is introduced. The preference weight is calculated according to the user's score on the item and the memory retention rate, and then the system coverage and click rate are improved by carefully merging the similarity values. Through comparative experiments with other link prediction algorithms on the dataset MovieLens, the results show that the hit rate of this algorithm is about 7% higher than that of other algorithms, and the coverage rate is slightly higher than that of the existing algorithms. |
| Keywords: recommendation algorithm; cosine similarity; interest drift; link prediction |
