摘要
协同过滤推荐算法具有实现容易、准确性高和效率高等优点,但也存在数据稀疏、冷启动和扩展性差等问题,并且在计算项目间的相似度时主要考虑项目评分相似,而对项目属性相似的考虑不够充分。为此,提出了一种改进计算项目属性相似度的算法,该算法可减小数据稀疏对推荐准确性的影响。根据项目被访问的情况,对项目的每个属性赋予不同的权重,通过对属性相似度的计算来提高对新加入项目的推荐度。实验结果表明该算法在推荐准确度上优于传统的协同过滤算法。
Collaborative filtering recommendation algorithm had the advantages of easy implementation,high accuracy and high efficiency,but it had the problems of sparse data,poor cold start and scalability.In calculating the similarity between projects,it mainly considered the similarity of project scores,instead of the similarity of project attributes.Therefore,an improved algorithm for calculating item attribute similarity was proposed,which could reduce the impact of data sparsity on the accuracy of recommendation.According to the situation that the project was visited,each attribute of the project was given different weights,and then the recommendation degree of the newly joined project was improved by calculating the similarity of the attributes.The experimental results have shown that the proposed algorithm outperforms the traditional collaborative filtering algorithm in recommendation accuracy.
作者
李转运
孙翠敏
LI Zhuanyun;SUN Cuimin(Computer Department,Anhui Post and Telecommunication College,Hefei 230031,China)
出处
《新乡学院学报》
2019年第3期30-33,共4页
Journal of Xinxiang University
关键词
推荐系统
协同过滤
属性权重
近邻模型
recommendation system
collaborative filtering
attribute weight
near neighbor model