摘要
为了提高推荐系统的可扩展性和用户满意度,设计基于数据挖掘和聚类分析的协同过滤推荐算法。基于双向关联规则原理,构建标签资源矩阵,利用K-means聚类算法对标签进行聚类。结合用户偏好标签,算法能计算标签与资源的紧密程度,实现基本推荐。通过标签计算用户与资源的兴趣度,实现个性化推荐。将基本推荐和个性化推荐线性组合,得出最终结果。实验表明,该算法不仅能保持数据集的平衡状态,准确性也高。通过聚类捕捉更复杂的用户兴趣模式,显著提高了推荐结果的命中率和NDCG值,为用户提供更符合个性化需求的资源。
In order to improve the scalability and user satisfaction of recommendation systems,a collaborative filtering recommendation algorithm based on data mining and clustering analysis is designed.Based on the principle of bidirectional association rules,construct a label resource matrix and use K-means clustering algorithm to cluster labels.By combining user preference tags,the algorithm can calculate the degree of tightness between tags and resources,and achieve basic recommendations.Calculate user and resource interests through tags to achieve personalized recommendations.Combine basic recommendations and personalized recommendations linearly to obtain the final result.The experiment shows that the algorithm can not only maintain the balance of the dataset,but also has high accuracy.By capturing more complex user interest patterns through clustering,the hit rate and NDCG value of recommendation results are significantly improved,providing users with resources that better meet personalized needs.
作者
何岫钰
HE Xiuyu(School of Business,Beijing Language and Culture University,Beijing 100083,China)
出处
《电子设计工程》
2024年第9期47-50,共4页
Electronic Design Engineering
关键词
数据挖掘
聚类分析
协同过滤推荐
标签相似度
偏好度
个性化推荐
data mining
cluster analysis
collaborative filtering recommendation
label similarity
pref-erence degree
personalized recommendation
