摘要
针对评分数据稀疏和单一评分相似性计算不准确导致推荐质量不高的问题,提出一种基于项目兴趣度的协同过滤新算法。该算法先预测未评分项的值,在此基础上基于项目的分类、评分值及次数引入项目—项目类别兴趣度特征向量计算同组项目的相似性,提高了项目间相似性计算的准确度和推荐质量,避免了大量无用计算,提高了算法的效率。实验结果表明,该算法对目标项目预测评分的准确性、推荐质量及效率更高。
Aiming at the problems of lower recommendation quality caused by rating data sparseness and computed single rat- ing similarity inaccurate in recommender system, this paper proposed a novel collaborative filtering algorithm based on interest degree of item. The algorithm first to predict item rating that users had not rated, and then to compute the same group items similarity depended on the interest degree feature vectors of item-item classification, which was defined by item categories, marks and rating times. This method can be better to avoid mass of useless similar computations and imprecise single-rating-similarity computing,which improve the accuracy of similarity, recommended quality and the efficiency of the algorithm. The experimental results show that the new algorithm can efficiently improve the predicted accuracy of not rated item, and provide better recommendation results and efficiency.
出处
《计算机应用研究》
CSCD
北大核心
2013年第12期3618-3621,共4页
Application Research of Computers
基金
河北省高等学校科学研究计划青年基金资助项目(Q2012110)
关键词
兴趣度特征向量
数据稀疏
项目相似性
推荐质量
协同过滤
interest degree feature vectors
data sparseness
item similarity
recommendation quality
collaborative filtering
作者简介
孙光明（1979-），男，湖北鄂州人，讲师，工学硕士，主要研究方向为协同过滤算法、组播路由算法（sungmez@gmail．com）；
王硕（1979-），女，河北威县人，讲师，主要研究方向为协同过滤算法．
