基于信任度的协同过滤组合新闻推荐算法  

Trust-based New Recommendation Algorithm of Collaborative Filtering Combination

作　　者：刘亚军[1,2] 鲍娌娜[2] 高莉莎[3] 

机构地区：[1]三江学院计算机科学与工程学院,江苏南京210012 [2]东南大学计算机网络和信息集成教育部重点实验室,江苏南京210096 [3]南京供电公司,江苏南京210019

出　　处：《信息工程期刊（中英文版）》2014年第6期147-151,共5页Scientific Journal of Information Engineering

基　　金：‘江苏省“十二五”一级学科重点建设学科新闻传播学建设项目系列研究成果

摘　　要：针对常见单一推荐技术产生的推荐结果准确率不高并存在一定局限性问题，本文将社会学中的信任关系引入到个性化新闻推荐中，提出了基于信任度的协同过滤组合新闻推荐算法。首先通过推荐的准确率度量用户对推荐算法的信任度，然后基于信任度将Item—based协同过滤推荐算法和User-based协同过滤推荐算法的推荐结果进行组合，以提高推荐的质量。实验结果表明基于信任度的组合推荐算法能更多的向用户推荐感兴趣的新闻，具有更好的推荐效果。Aimed at resulting accuracy for common single recommendation technology are not high and there are some limitations, trust relationship in sociology is introduced into personalized news recommendation. This article raises a trust-based new recommendation algorithm of collaborative filtering combination. First, it measures users＇ trust to recommendation algorithm through the recommendation accuracy; then, in order to raise quality, the recommendation results of user-based and item-based collaborative filtering recommendation algorithm are combined through the trust. The experimental results show that trust-based combining recommendation algorithm have more accurately predict the degree of user interest in news than user-based and item-based collaborative filtering recommendation algorithm and more news of interest to recommend to the user, so it has the effect of better recommended

关 键 词：信任度 协同过滤 新闻推荐 组合推荐算法 

分 类 号：TP[自动化与计算机技术]