基于在线评论的顾客满意度评估方法

Evaluation Method of Customer Satisfaction Based on Online Reviews

基于在线评论对顾客满意度进行评估,并分析顾客满意度与评估属性的动态关系。首先,基于LDA从在线评论中提取评估属性,利用IOVO-SVM进行情感分析并计算各评估属性在不同时期的情感得分。然后,基于离差最大化方法确定各评估属性的权重。在此基础上,利用基于马氏距离的TOPSIS方法评估不同时期的顾客满意度。进一步地,通过构建向量自回归模型,分析顾客满意度与各评估属性的动态关系。最后,以一家三星酒店为例,说明该方法的可行性和有效性。本文提出的方法可以帮助管理者评估顾客满意度,并了解各评估属性对顾客满意度影响的动态变化及差异,进而辅助企业进行有针对性的产品或服务改进以提高顾客满意度。

Improving customer satisfaction is a top priority for businesses.In the traditional customer satisfaction evaluation methods,information to assess customer satisfaction is mainly obtained through questionnaires,surveys,and interviews.However,they have the disadvantage of being time-consuming,labor-and material-intensive investments,and the data is easily out of date.For the past years,with the rapid development of the Internet and e-commerce,more and more e-commerce sites have allowed consumers to post online reviews about their experiences with a product or service,which are characterized by large sample data,easy accessibility,high authenticity,low cost and dynamic updates.As a result,online reviews have become an important source of information for companies to analyze the customer satisfaction of products or services.Although existing studies on customer satisfaction assessment and influence identification based on online reviews have achieved results,there are still limitations.First,most of the current customer satisfaction assessments based on online reviews assume that the evaluation attributes are independent of each other,and less consideration is given to possible correlation between the evaluation attributes.Second,few studies have analyzed the dynamic relationship between customer satisfaction and evaluation attributes,but through dynamic relationship analysis,we can more intuitively understand the dynamic changes in the impact of each evaluation attribute on customer satisfaction.In this paper,we evaluate customer satisfaction through online reviews and analyze the dynamic relationship between customer satisfaction and evaluation criteria.First,we extract the evaluation criteria from the online reviews based on LDA,use IOVO-SVM to identify sentiment orientation with respect to the criteria,and calculate sentiment scores for the criteria in different time periods.The weights of the criteria are then determined based on the maximum deviation method.In this paper,we use the TOPSIS method with Mahalanobis distance to evaluate customer satisfaction in different time periods.Furthermore,based on the customer satisfaction evaluation values and the sentiment scores of each evaluation attribute in different periods,the dynamic relationship between customer satisfaction and each evaluation attribute is analyzed by constructing the Vector Autoregressive(VAR)model.Finally,a three-star hotel is used to illustrate the process and effectiveness of the method.The data comes from Qunar.com,where we collect 1,850 online reviews of VIH hotel from January 2015 to July 2019.In the case study in this article,one month is used as a time period,so a total of 55 time periods is included.The results show that VIH hotel price has the fastest impact on customer satisfaction and the largest contribution rate to improving customer satisfaction compared to other evaluated attributes.VIH hotel can focus on price to improve customer satisfaction,followed by improvements in rooms,cleanliness and cost performance.The proposed approach can help managers evaluate customer satisfaction and understand the dynamic changes and differences in the impact of each evaluation criterion on customer satisfaction,which can then assist companies in improving customer satisfaction through targeted product or service improvements.There are some limitations to this study that may serve as avenues for future research.Only customer-generated online review data is used in this study.Correspondingly,a large amount of customer-generated rating data is publicly available on the website,which contains a wealth of valuable information about customer satisfaction.Thus,how to analyze customer satisfaction more deeply based on these ratings is a promising direction.