融合社交网络和分层自注意力的团购推荐算法  

Group buying recommendation algorithm based on social network and hierarchical self attention mechanism

作　　者：孙男男 朴春慧 马新娜[1,2] SUN Nan-nan;PIAO Chun-hui;MA Xin-na(School of Information Science and Technology,Shijiazhuang Tiedao University,Shijiazhuang 050043,China;Key Laboratory for Electromagnetic Environmental Effects and Information Processing,Shijiazhuang Tiedao University,Shijiazhuang 050043,China)

机构地区：[1]石家庄铁道大学信息科学与技术学院,河北石家庄050043 [2]石家庄铁道大学河北省电磁环境效应与信息处理重点实验室,河北石家庄050043

出　　处：《计算机工程与设计》2023年第6期1867-1874,共8页Computer Engineering and Design

基　　金：国家自然科学基金项目(12172234)。

摘　　要：针对团购群组推荐中预定义的融合策略不能动态获取群组用户权重,以及群组-项目交互数据的稀疏性等问题,提出融合社交网络和分层自注意力的团购推荐算法SSAGR。采用循环神经网络,捕捉团购中用户随时间变化的复杂潜在兴趣;利用用户级和群组级自注意力网络,将社交网络信息整合到用户表示中,实现不同权重下的群组偏好聚合策略;通过神经协同过滤NCF,挖掘群组-项目交互,完成团购推荐。在MaFengWo和Douban Book数据集上的实验结果表明,SSAGR方法相较对比算法具有更好的群组推荐效果。Aiming at the problems that the predefined fusion strategy cannot reflect the dynamic weight of the group user,and the group-item interaction data are sparse,a group buying recommendation model based on social network and hierarchical self-attention network(SSAGR)was proposed.Recurrent neural network was used to capture the complex potential interests of users in group buying over time.Using user-level and group-level self-attention networks,the user’s social network information was integrated into the user representation,and a group preference aggregation strategy under different weights was implemented.Through neural collaborative filtering(NCF),the group-item interaction was mined to complete the group buying recommendation.Experimental results on the MaFengWo and Douban Book datasets show that the SSAGR method has better group recommendation effects than the comparison algorithms.

关 键 词：团购 群组推荐 融合策略 社交网络 循环神经网络 自注意力网络 神经协同过滤 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]