MOOB:一种改进的基于Bandit模型的推荐算法
MOOB: An Improved Recommendation Algorithm Based on Bandit Model
投稿时间:2017-10-24  修订日期:2017-10-24
DOI:
中文关键词: Bandit模型  马太效应  长尾现象  多目标优化  覆盖率
英文关键词: Bandit model  Matthew effect  long tail phenomenon  multi-objective optimization  coverage
基金项目:国家科技支撑计划项目子课题(2015BAD29B01),中央高校基本科研业务费专项资金项目(CZP17007)
作者单位E-mail
帖军 中南民族大学计算机科学学院 tiejun@mail.scuec.edu.cn 
孙荣苑 中南民族大学  
孙翀 中南民族大学  
郑禄 中南民族大学  
摘要点击次数: 15
全文下载次数: 
中文摘要:
      传统Bandit模型在推荐系统应用中易产生马太效应和长尾现象,本文提出一种基于置信区间上界算法的多目标优化推荐算法。该算法可以在保证预测精准度的基础上有效地避免马太效应,并提高推荐系统对长尾物品的挖掘能力。论文采用YaHoo的新闻推荐数据集对算法进行实验和评价,实验结果表明多目标优化推荐算法能够在预测准确率较高的情况下,有效地解决长尾物品发掘问题,避免马太效应,提高推荐系统的精度和广度。
英文摘要:
      Traditional Bandit model is easy to generate Matthew effect and long tail phenomenon in recommendation system. This paper proposes a multi-objective optimization recommendation algorithm based on confidence interval upper bound algorithm. The algorithm can effectively avoid Matthew effect on the basis of ensuring the accuracy of prediction, and improve the mining ability of the recommendation system to the long tail items. The paper uses YaHoo's news recommendation data set to experiment and evaluate the algorithm, experimental results show that the multi - objective optimization recommendation algorithm can effectively solve the problem of long - tail item excavation, avoid the Matthew effect and improve the precision and breadth of the recommended system under the condition of high prediction accuracy.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭