摘要
在传统协同过滤算法中,相似度直接依据用户评分。但是,用户评分会受各种不确定因素影响。采用数值评分的推荐系统收集到的用户喜好信息是模糊、不精确和不完整的。单一的数值不能包含丰富的信息来表达用户喜好,也会导致推荐结果的不准确性。文中定义了几种模糊集的隶属函数,提出了基于模糊逻辑的相似度计算方法。实验结果表明,基于模糊权重的相似度有效的提高了推荐系统的预测准确度,一定程度上解决了协同过滤算法的可扩展性和数据稀疏性问题。
In the traditional collaborative filtering algorithm,the calculation of similarity is based directly on user ratings,which are subject to uncertain factors,and thus the user preferences information is inaccurate by Numerical rating. This paper defines several membership functions of fuzzy sets and puts forward the similarity calculation method based on fuzzy logic. The experimental results show that the similarity based on fuzzy weight effectively improves the accuracy of the recommendation system.
出处
《电子科技》
2015年第7期111-114,共4页
Electronic Science and Technology
关键词
推荐系统
协同过滤
相似度
模糊权重
recommendation system
collaborative filtering
similarity
fuzzy weight
