摘要
协同过滤中的评分数据稀疏性使得最近邻搜寻不够准确,导致推荐质量较差.基于此,文中提出一种有效的针对稀疏评分的最近邻选择方法——两阶段最近邻选择算法(TPNS).TPNS分为两个步骤,首先计算用户间的近邻倾向性,选择近邻倾向性较高的用户组成初始近邻集合;然后根据初始近邻集合计算目标用户与其他用户间的等价关系相似性,使用等价关系相似性对目标用户的初始近邻集合进行修正,得到最近邻集合.在MovieLens数据集上对比常用的推荐算法,实验结果表明文中方法在协同过滤推荐的应用中具有更高的准确性.
In collaborative filtering, sparsity in ratings resuhing in poor recommendations. To address this makes inaccurate neighborhood formation, thereby issue, a method of neighborhood formation, two-phase neighbor selection method （TPNS）, is proposed. The definition of neighbor tendency is given. Based on the neighbor tendency, the preliminary neighborhood is formed. Then, the equivalence relation similarity is applied to modify the preliminary neighborhood, which makes the neighborhood formation more accurate. Experimental results on MovieLens dataset show that compared with the existing algorithms, TPNS performs better in the application of personalized recommendation.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2013年第10期968-974,共7页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金(No.71271072
71201145)
高等学校博士学科点专项科研基金(No.20110111110006)
教育部人文社会科学研究基金(No.09YJC630055
11YJC630283)资助项目
关键词
推荐系统
协同过滤
最近邻选择
近邻倾向性
近邻修正
Recommender System, Collaborative Filtering, Neighborhood Formation, Neighbor Tendency, Neighborhood Modification
