摘要
用户真实的购买场景中,购物不仅仅看兴趣,当下以及未来需求也很重要,而现有大部分推荐方法研究的是挖掘用户近期兴趣,较少从商品间的关系来研究用户潜在需求.为了提高推荐算法准确性,丰富推荐种类,本文将商品互补替代关系特征和购买先后序列模式融入到推荐算法中,提出一种考虑商品互补替代关系和购买序列模式来研究用户潜在需求的推荐算法,该算法在亚马逊公开数据集Grocery上进行验证,并与相关算法进行对比,结果表明所提算法在命中率HR(hits ratio)和归一化折损累计增益NDCG(normalized discounted cumulative gain)指标上均得到有效的改进.
In the real purchasing scenario of users,shopping is not only based on interests,but also on current and future needs.However,most of the existing recommendation methods focus on mining users recent interests,and rarely study users potential needs from the relationship between products.In order to improve the accuracy of the recommendation algorithm and enrich the types of recommendation,this paper integrates the characteristics of commodity complementary substitution relationship and purchase sequence pattern into the recommendation algorithm,and proposes a recommendation algorithm that considers the commodity complementary substitution relationship and purchase sequence pattern to study the potential needs of users.The algorithm is verified on Amazon public data set Grocery.Compared with the relevant algorithms,the results show that the proposed algorithm is effectively improved in hits Ratio HR(hits ratio)and normalized discounted cumulative gain(NDCG).
作者
任志波
戎秀玲
宋欣欣
REN Zhibo;RONG Xiuling;SONG Xinxin(Comprehensive Experiental Center,Hebei University,Baoding 071002,China;Dancing College,Xinjiang Arts University,Urumchi 830000,China)
出处
《河北大学学报（自然科学版）》
CAS
北大核心
2024年第5期551-560,共10页
Journal of Hebei University(Natural Science Edition)
基金
河北省社会科学基金资助项目(HB19Q011)。
关键词
互补搭配
序列模式
个性化推荐
complementary pairing
sequence mode
personalized recommendations
作者简介
第一作者:任志波(1971-),男,河北大学教授,博士,主要从事个性化推荐算法、数据挖掘方向研究.E-mail:renzhibo@hbu.edu.cn;通信作者:宋欣欣(1998-),女,新疆艺术学院助教,主要从事个性化推荐算法方向研究.E-mail:1244879013@qq.com。
