首页> 外文会议>IEEE Global Conference on Consumer Electronics >Negative confidence recommendation system avoids local solution based on user’s negative reactions
【24h】

Negative confidence recommendation system avoids local solution based on user’s negative reactions

机译:负信信推荐系统避免了基于用户的负反应的本地解决方案

获取原文

摘要

There are situations that one cannot get enough negative data for the machine learning. We show that if one can equip data with confidence values, one can successfully learn even there are only limited negative data. We numerically illustrate the behaviour of our proposed method on synthetic datasets firstly. The accuracy of the proposed method is 24.4% and 10.1% higher than two conventional methods on average. We further conducted movie recommendation experiment using MovieLens dataset to show the usefulness of the proposed method. Experimental results show that 33.6 % of previously-watched movies are reduced from the recommendation and genre-rating value is 4.67% higher in our proposed method.
机译:有些情况下,一个人无法获得机器学习的足够负数据。我们表明,如果可以用置信度值装备数据,即使只有有限的否定数据也可以成功学习。我们首先说明了我们在合成数据集上提出的方法的行为。该方法的准确性为24.4%,平均常规方法高出10.1%。我们进一步使用Movielens DataSet进行电影推荐实验,以显示所提出的方法的有用性。实验结果表明,33.6%的先前观看的电影从建议书中减少了我们所提出的方法的推荐和类型额定值的4.67%。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号