西安电子科技大学学报

• 研究论文 • 上一篇    下一篇

长尾群组推荐的免疫多目标优化实现

韩亚敏1;柴争义1,2,3;李亚伦1;朱思峰3   

  1. (1. 天津工业大学 计算机科学与软件学院,天津 300387;
    2. 诺丁汉大学(英国) 计算机学院,诺丁汉 NG8 1BB;
    3. 周口师范学院 数学与统计学院,河南 周口 466001)
  • 收稿日期:2017-06-29 出版日期:2018-06-20 发布日期:2018-07-18
  • 通讯作者: 柴争义(1976-), 男, 副教授, E-mail: super_chai@126.com
  • 作者简介:韩亚敏(1993-), 女, 天津工业大学硕士研究生, E-mail: han_min8013@163.com
  • 基金资助:

    国家自然科学基金资助项目(U1504613)

Immune multiobjective optimization on long tail group recommendation

HAN Yamin1;CHAI Zhengyi1,2,3;LI Yalun1;ZHU Sifeng3   

  1. (1. School of Computer Science & Software Engineering, Tianjin Polytechnic Univ., Tianjin 300387, China;
    2. School of Computer Science, Univ. of Nottingham, Nottingham NG8 1BB, UK;
    3. School of Mathematics & Statistics, Zhoukou Normal Univ., Zhoukou 466001, China)
  • Received:2017-06-29 Online:2018-06-20 Published:2018-07-18

摘要:

针对群组推荐仅关注推荐准确度的不足, 提出一种引入长尾物品的群组推荐算法.由于长尾推荐会降低推荐系统的准确度, 算法将群组推荐建模成多目标问题, 以群组用户满意度和物品流行度为目标函数, 设计适合多目标推荐问题的编码、交叉、变异等免疫算子, 采用免疫多目标算法优化长尾群组推荐.在公开数据集上的实验结果表明, 所提算法在保持群组推荐准确度的同时, 提高了推荐结果的多样性和新颖性.

关键词: 推荐系统, 多目标优化, 免疫算法, 长尾理论

Abstract:

Traditional group recommendations pay more attention to accuracy, which ignores the importance of long tail items. Considering the above problem, a long tail group recommendation algorithm is proposed. However, long tail recommendations reduce the accuracy of recommendation systems. In this paper, the group recommendation is modeled as a multiobjective problem. The group user satisfaction and item popularity are used as objective functions, and the immune operators such as coding, crossover and mutation are designed for multiobjective recommendation. Then, the long tail group recommendation is optimized by the immune multiobjective algorithm. Experimental results show that the proposed algorithm improves the diversity and novelty of the recommendation results while maintaining the accuracy of the group recommendation.

Key words: recommendation systems, multiobjective optimization, immune algorithm, long tail theory