摘要
互联网的迅速发展,将人们带入了信息爆炸的时代,而个性化推荐系统是解决该问题的一种非常有效的方法.研究的Slope One算法是一种基于项目的协同过滤推荐算法,该方法简单、高效、易于实现且具有合理的准确性.然而,该算法在偏差计算时考虑了所有共同评分的项目,在某些情况下,对不相关项目计算偏差可能会降低预测的准确性.为了在计算偏差时过滤掉不相关项目对预测结果的影响,将项目分类和K近邻引入Slope One算法,以期得到更好的准确性.最后,在MovieLens数据集上的实验结果表明,提出的方法在数据稀疏和共同评分项目较少的情况下仍能得到较好的准确性.
With the rapid development of the Internet,it brings people into the era of information explosion,and the personalized recommendation system is a very effective way to solve this problem.In this paper,the Slope One algorithm is a kind of item-based collaborative filtering recommendation algorithm,and it is simple,efficient,easy to achieve,and it has reasonable accuracy.this algorithm takes all items into consideration when computing deviation,however,in some situation,computing deviation between unrelated items may finally decrease the prediction accuracy.Thus,to avoid causing a bad influence on the predicted results,we should filter unrelated items when computing deviation.Finally,we make experiments on the MovieLens datasets,and results show that the proposed algorithm in the case of sparse datasets and less neighbor achieves better prediction accuracy.
出处
《小型微型计算机系统》
CSCD
北大核心
2017年第9期2090-2095,共6页
Journal of Chinese Computer Systems
基金
国家自然科学青年基金项目(51305383)资助
教育部博士点专项基金项目(20131333120007)资助
关键词
个性化推荐
协同过滤
项目分类
项目K近邻
SLOPE
One算法
personalized recommendation
collaborative filtering
item category
Knearest neighbor of item
Slope One algorithm
