欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
在信息过载时代,推荐系统能够帮助用户发现感兴趣的内容。协同过滤是推荐系统中最常用的技术,然而传统的协同过滤算法未能充分考虑项目标签对相似度的影响,因而推荐质量不高。文中提出了一种结合用户评分和项目标签的协同过滤算法,算法中关键的相似度计算是对评分相似度和标签相似度的加权,通过加权降低了相似度矩阵的稀疏性,并且保证项目之间只有在共同评分较多且标签相似时才具有较高的相似度,从而使相似度计算更加准确。通过对比实验得出加权系数在0.3-0.5时推荐质量较高,在公开数据集上与传统协同过滤算法的比较结果表明,文中的算法在平均绝对误差上降低了约3%。
英文摘要：
Recommender system can help users find their interests in the era of information overload. Collaborative filtering is one of the most widely used techniques,while the traditional collaborative filtering algorithm has a low er effectiveness because of the seldom consideration about the similarity of item tags. A collaborative filtering algorithm combined ratings with tags is proposed in this paper,the most important procedure of the algorithm is to calculate the similarity of ratings and tags. The tw o kinds of similarity is combined by weighting,through this method the sparsity of the similarity matrix is greatly reduced and ensures that only when the items have more common ratings and in the same tags can reach a high similarity,so that the accuracy of the calculation of similarity is highly increased. The optimal range of the weighting coefficient is 0. 3 to 0. 5 by comparative experiment. Compared with the traditional collaborative filtering algorithm on public data set,the algorithm proposed in this paper improves the accuracy of recommendation to 3% in MAE.
位置：
同期刊论文项目
