面向在线评论的用户需求分析框架与实证研究--基于KANO模型

Analytical Framework and Empirical Study of User Needs for Online Reviews Based on KANO Model

[目的/意义]理解用户需求是企业管理实践中的重要一环,在线评论作为用户偏好表达的重要渠道,是企业发现用户需求的重要数据来源。因此,如何从在线评论中有效挖掘出用户需求,是一个重要的研究问题。[方法/过程]文章构建了一个基于KANO模型、面向在线评论的用户需求分析框架。首先利用LDA主题模型从在线评论中抽取商品属性,然后基于BERT模型标注出不同属性上的评论短句,并对其进行情感分析,最后依据不同属性上的评论短句与情感分析结果,通过KANO模型分析用户属性需求。以手机评论数据为例对模型进行验证。[结果/结论]将商品属性分为必备属性、期望属性、魅力属性与无差异属性四类,并以可视化结果呈现。实验结果表明面向在线评论的用户需求分析框架能够有效提取在线评论中用户的属性需求信息,为企业提供产品改进策略参考。[局限]样本数据集有限,存在不均衡状况。

[Purpose/significance]Practice of management requires a deep understanding of user needs.In addition,as a significant channel of user preference expression,online review is promising source from which to reveal user needs.Therefore,how to identify user needs from massive online reviews become an important issue.[Method/process]In this case,based on KANO model,we propose an analytical framework of user needs for online reviews.Specifically,product attributes are first elicited from online reviews based on LDA.Afterwards,the short sentences are labeled into different attributes using BERT model,and the sentiment orientations of the short sentences are identified at the same time.Naturally,user needs can be identified by KANO model according to the short sentences and sentiment analysis.Besides,a cell phone review dataset is used to validate the proposed model.[Result/conclusion]The framework divides product attributes into four categories:must-be quality attribute,one-dimensional quality attribute,attractive quality attribute,and indifferent quality attribute,which will be presented in visualization.The empirical study show that user needs can be effectively identified,and improvement strategies can be provided for enterprises.[Limitations]The dataset in this paper is limited and unbalanced.