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：
为了解决电子商务推荐系统在推荐新项目方面的冷启动问题，同时提高用户与推荐项目的相似度，通过对比当前的推荐算法，提出一种结合可信度和动态时间加权的推荐算法。该算法引入用户评分可信度来计算用户和项目的相似性，将新项目推荐给可信度高的用户；分析用户兴趣、项目受欢迎度和时间的关系构造动态时间加权函数，将项目推荐给用户兴趣度高且项目受欢迎度高的用户。通过实验验证该算法与传统的基于用户的推荐UBCF算法相比能够提高近7％的推荐准确度，与基于项目的推荐IBCF算法相比能够提高近4．7％的推荐准确度．同时解决新项目推荐的冷启动问题。
英文摘要：
To solve cold start problem which the new project recommends in e-commerce recommendation system and improve the similarity of us- er-user and item-item. Proposes a recommendation algorithm combining score＇s credibility and dynamic time weighted by contrasting the current recommendation algorithm. The proposed method introduces the credibility of users＂ ratings to compute the similarity of user-user and item-item. Then the new items are recommended to the user of high credibility. Dynamic time weighted function is constructed by analyzing the relationship between time and users＂ interests or the popularity of the project. Then the items are recommended to the user of high interests and the item of high popularity. The algorithm is verified by experiment that it compared with the traditional user-based collaborative filtering UBCF algorithm can improve the accuracy of nearly 7% of the recommended, and it compared with item-based col- laborative filtering IBCF algorithm can improve the accuracy of nearly 4.7% of the recommended. At the same time, the algorithm solves the problem of cold start in recommendation of new project.
同期刊论文项目
同项目期刊论文
期刊信息
位置：