基于类别偏好Canopy-K-means的协同过滤推荐系统算法

邹燕飞

基于类别偏好Canopy-K-means的协同过滤推荐系统算法

投稿时间：2019-11-17 修订日期：2019-12-10 点此下载全文

引用本文：

摘要点击次数: 259

全文下载次数: 0

作者	单位	邮编
邹燕飞^*	咸阳师范学院计算机学院	712000

中文摘要:协作过滤算法(CF)在推荐系统中难以处理数据的稀疏性和可伸缩性问题。本文提出了基于类别偏好Canopy-K-means的协同过滤算法(CPCKCF),设计了用户项类别偏好比率(UICPR)的定义,并用来计算UICPR矩阵。将Canopy算法作为CPCKCF的前置算法,并将输出作为K-means算法的输入,其结果用于用户数据进行聚类并找到最近的用户以获得预测得分,使用MovieLens数据集进行的实验结果表明,与传统的基于用户的协作过滤算法相比,所提出的CPCKCF算法将计算效率和推荐精度提高了2.81%。

中文关键词:推荐系统协同过滤数据挖掘 Canopy算法 K-means算法

Algorithm of Collaborative Filtering Recommendation System based on Category Preference Canopy-k-means

Abstract:Collaborative filtering algorithm (CF) is difficult to deal with data sparsity and scalability in recommendation system. This paper proposes a collaborative filtering algorithm (cpckcf) based on category preference canopy-k-means, designs the definition of user item category preference ratio (uicpr), and uses it to calculate uicpr matrix. Canopy algorithm is used as the pre algorithm of cpckcf, and the output is used as the input of K-means algorithm. The results are used to cluster the user data and find the nearest user to obtain the prediction score. The experimental results using movielens data set show that the proposed cpckcf algorithm improves the calculation efficiency and recommendation accuracy by 2.81% compared with the traditional user based collaborative filtering algorithm.

keywords:recommendation system collaborative filtering data mining canopy algorithm k-means algorithm

查看全文 查看/发表评论 下载pdf阅读器