摘要
提出一种端到端的基于产品方面的神经网络推荐模型。该模型利用产品方面标签注意力机制,建模了用户偏好和项目特性之间的联系,并对用户和项目采用方面级别的表示,模拟用户与项目间的细粒度交互过程,从而获得更精确和更具解释性的推荐结果。在COAE中文汽车领域数据集和Yelp基准数据集上分别进行实验,结果表明,所提模型的性能明显优于ANR和NARRE模型。
An end-to-end neural network recommendation model based on product aspect was proposed.The relationship between user preferences and item features was modeled by using aspect labels attention in the model.More precise and explanatory recommendation results were obtained by modeling the fine-grained interaction process between user and item under the aspect level representation of user and item.The experimental results showed that the performance of the proposed model significantly outperformed ANR and NARRE models on COAE Chinese automobile field data set and Yelp benchmark data set.
作者
王素格
刘宇飞
李旸
符玉杰
郑建兴
WANG Suge;LIU Yufei;LI Yang;FU Yujie;ZHENG Jianxing(School of Computer and Information Technology, Shanxi University, Taiyuan 030006, China;Key Laboratory of Computational Intelligence and Chinese Information Processing of Ministry of Education, Shanxi University, Taiyuan 030006, China;School of Finance, Shanxi University of Finance and Economics, Taiyuan 030006, China)
出处
《郑州大学学报（理学版）》
北大核心
2022年第1期48-53,共6页
Journal of Zhengzhou University:Natural Science Edition
基金
国家自然科学基金项目(62076158,62072294)
山西省重点研发计划项目(201803D421024)。
关键词
推荐模型
文本评论
细粒度
方面标签
recommendation model
text review
fine-grained
aspect label
