作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 人工智能及识别技术 • 上一篇    下一篇

基于多维相似度的利基产品推荐方法

刘业政,熊强,姜元春   

  1. (合肥工业大学 管理学院,合肥 230009)
  • 收稿日期:2016-12-08 出版日期:2018-03-15 发布日期:2018-03-15
  • 作者简介:刘业政(1965—),男,教授、博士、博士生导师,主研方向为数据挖掘、机器学习、电子商务;熊强,博士研究生;姜元春,副教授、博士。
  • 基金资助:
    国家自然科学基金重大项目(71490725);国家自然科学基金(91546114,71501057);国家科技支撑计划项目“第三方检验检测科技服务云平台研发及示范应用”(2015BAH26F00)。

Recommendation Method for Niche Product Based on Multi-dimensional Similarity

LIU Yezheng,XIONG Qiang,JIANG Yuanchun   

  1. (School of Management,Hefei University of Technology,Hefei 230009, China)
  • Received:2016-12-08 Online:2018-03-15 Published:2018-03-15

摘要: 电子商务平台上的产品销售具有长尾特征,但现有以追求精度为目标的推荐方法难以将处于长尾上的利基产品加入推荐列表。为此,从利基产品视角出发提出一种新的推荐方法。基于用户评分、产品属性和隐特征信息分别计算用户之间的评分相似度、偏好相似度和隐特征相似度,并综合这三种相似度挖掘利基产品高评分用户的相似用户,从而得到利基产品的受众并为其进行推荐。实验结果表明,该方法针对利基产品的推荐转化率远高于概率矩阵分解和协同过滤方法,在解决利基产品推荐问题上更有效。

关键词: 推荐系统, 长尾产品, 利基产品, 相似度计算, 受众

Abstract: Sales of the e-commerce platform possess a long tail character and niche products in the long tail are difficult to be involved in the list produced by the recommendation method whose goal is the pursuit of precision.Aiming at this problem,from the perspective of niche product,this paper proposes a new recommendation method.It calculates user ratings similarity,preferences similarity and latent features similarity between users based on rating information,attribute information and latent feature information respectively.Then,it excavates the possible users of the niche products that have the top similarity with the users who have high ratings for niche products based on the three similarities,and provids niche products for those possible users.Experimental results show that the recommendation conversion rate of the proposed method is much higher than probability matrix factorization method and collaborative filtering method for niche product recommendation.Therefore,it is more effective to solve the problem of niche products recommendation.

Key words: recommender system, long tail product, niche product, similarity calculation, possible user

中图分类号: