| 石宇,查梦娟,梁宇,邱惠.基于群体兴趣的个性化推荐候选集构建[J].数字图书馆论坛,2017,(1):51~56 |
| 基于群体兴趣的个性化推荐候选集构建 |
| Construction of Candidate Item Sets in Personalized Recommendation Based on Group Interest |
| |
| DOI： |
| 中文关键词: 个性化推荐;群体兴趣;候选集 |
| 英文关键词: Personalized Recommendation;Group Interest;Candidate Item Sets |
| 基金项目: |
| 作者 | 单位 | | 石宇 | 武汉大学 | | 查梦娟 | 武汉大学 | | 梁宇 | 国网湖北省电力公司鄂州供电公司 | | 邱惠 | 国网湖北省电力公司鄂州供电公司 |
|
| 摘要点击次数: 1886 |
| 全文下载次数: 1289 |
| 中文摘要: |
| 对个性化推荐候选集构建方案进行优化,在保证对用户感兴趣项目覆盖率较高的前提下,压缩候选集规模,从而改善个性化推荐效果.采用融合用户群体兴趣与个性化特征的方法构建推荐候选集,先以用户群体兴趣为基础构建初始候选集,在此基础上基于用户个性化特征对其进行过滤,生成推荐候选集.在模型构建的基础上,以电影数据为例进行模型效果验证.实验结果显示,该模型在覆盖率小幅降低的情况下,可将候选集规模压缩为原来的3.4％;该模型能显著提升个性化推荐的最终效果,因此该策略是一种较理想的候选集构建策略. |
| 英文摘要: |
| The purpose of this paper is to optimize the construction method of candidate item sets.We improve the effect of personalized recommendation by reducing the number of candidate item sets without sacrificing the coverage of items that users interested.The construction of candidate item sets in the personalized recommendation is based on the method of fusing group interest and the personalized features.First of all,this paper consmucts the initial candidate item sets based on the group interest,then filter the candidate item sets based on user's personalized features.This paper verifies the effectiveness of the method by taking movie data as an example.The results show that the number of candidate item sets reduce to 3.4％ of the original when the coverage of items decrease slightly.Moreover,this method can improve the effect of personalized recommendation significantly.Thus,it can be concluded that it is an ideal tactic to construct the candidate item sets. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
