基于用户评论信息抽取的电商问答研究  

Research on E-Commerce Q&A Based on User Comment Information Extraction

作　　者：胡志强 钱宇[1] 袁华[1] 汪子牧 HU Zhiqiang;QIAN Yu;YUAN Hua;WANG Zimu(School of Management and Economics,University of Electronic Science and Technology of China,Chengdu 611731,China)

机构地区：[1]电子科技大学经济与管理学院,四川成都611731

出　　处：《信息系统学报》2021年第2期13-25,共13页China Journal of Information Systems

基　　金：国家自然科学基金项目(71572029,91846105,71671027)。

摘　　要：消费者在线购物时既面临着信息的不对称性,又面临着海量评论数据中包含着的丰富信息。本文提出一种基于注意力机制的评论检索模型,用于解决消费者在线购买决策过程中面临信息过载以及向已购买者提问却不能获得及时性回复的管理问题。该模型通过对已有的消费者在线评论进行检索,将候选答案推荐给相关问题的提问者。然而,由于用户提问与评论之间缺乏直接的对应关系,本文引入问题答案对已有的用户提问与评论进行匹配以构建训练样本从而提升匹配效果。实验结果表明本文所提出的基于注意力机制的电商评论检索模型具有较好的实用性。Consumers are not only faced with information asymmetry when shopping online,but also with abundant information contained in massive review data.A comment retrieval model based on attention mechanism is proposed in this paper to solve the information overload faced by consumers in the process of online purchase decision-making and the management problem of questioning the purchasers but unable to obtain timely responses.The model retrieves existing online consumer reviews and recommends candidate answers to questioners.However,since there is no direct correspondence between user questions and comments,this paper introduces question answers to match existing user questions and comments to build training samples and improve the matching effect.The experimental results show that the proposed attention-based comment retrieval model has good practicability.

关 键 词：问答任务 电商问答系统 用户评论 注意力机制 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]