APP下载

基于个性化特征的协同过滤推荐算法

2017-04-01李涛

现代电子技术 2017年5期

李涛

摘 要: 针对相似度特征点推荐方法对用户个性化需求匹配度不高的问题,提出基于个性化特征的协同过滤推荐算法。以社会网络为结构模型构建用户信息的评分模型和项目属性模型,采用信任度条件概率分析方法构建可靠性推荐模型,进行个性化特征分析和提取,实现个性化特征需求与项目兴趣点的合理匹配,实现协同过滤推荐,最后通过仿真实验进行测试分析。结果表明,采用该方法进行社会网络项目协同过滤推荐的用户评分高,平均绝对误差和均方根误差小,提升了推荐质量。

关键词: 个性化特征; 协同过滤推荐; 评分模型; 项目属性

中图分类号: TN911?34; TP391 文献标识码: A 文章编号: 1004?373X(2017)05?0078?04

Abstract: Since the recommendation method of the similarity feature point has low matching degree for the user′s persona?lized demand, a collaborative filtering recommendation algorithm based on personalized feature is put forward. The scoring model and project attribute model of the user information were constructed by taking the social network as the structure model. The trust degree conditional probability analysis method is adopted to construct the reliability recommendation mode. And then the personalized features are analyzed and extracted to match the personalized feature demand and project interest point reasonably, so as to implement the collaborative filtering recommendation. The test analysis was conducted with simulation experiments. The test results show that the method has high user scoring to perform the collaborative filtering recommendation of the social network project, the mean absolute error and root mean square error are small, and the recommendation quality is improved.

Keywords: personalized feature; collaborative filtering recommendation; scoring model; project attribute

0 引 言

社會网络为网络用户提供了个性化的定制需求,通过社会网络的个性化推荐系统,根据用户的兴趣偏好进行项目推荐,构建用户?项目的个性化推荐模型,在提高用户网络浏览的效率和满意度方面具有重要意义[1]。另一方面,通过制定个性化协同过滤推荐模型,将提高网络信息的推送度,提高网络项目的收益,研究社会网络的用户协同过滤推荐模型具有重要意义[2]。

网络的协同过滤推荐是以网络用户信任关系为基础的。通过对网络用户先验信息的浏览和评估,制定个性化需求,综合考虑社会网络结构和用户需求,制定个性化服务项目实现信息推送和网页推荐,为更高效率的网络体验奠定基础[3]。当前,典型的协同过滤推荐方法主……

登录APP查看全文