基于自然语言处理的车企客户反馈数据挖掘:上汽通用五菱公司的案例  

Customer Feedback Data Mining in Automobile Industry Based on Natural Language Processing:The Case of SAIC GM Wuling

作　　者：胡洁 练朝春 李艳婷[2] 雷月霆 岳子桐 HU Jie;LIAN Chaochun;LI Yanting;LEI Yueting;YUE Zitong(Shanghai GM Wuling Automobile Co.,Ltd.,Liuzhou,Guangxi 545000,China;School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China)

机构地区：[1]上海通用五菱汽车股份有限公司,广西柳州545000 [2]上海交通大学机械与动力工程学院,上海200240

出　　处：《工业工程与管理》2022年第6期192-200,共9页Industrial Engineering and Management

基　　金：国家自然科学基金面上项目(72072114)。

摘　　要：文本数据作为典型的非结构化数据,存在于汽车企业的全业务流程各个环节中。本文以上汽通用五菱公司“用户之声”数据集为背景进行案例研究,首先将文本数据进行预处理并构建汽车领域专业词典,然后将利用词向量训练模型获得的语义特征输入云算法池中的机器学习模型进行预测和分类,最终服务于应用层业务环节的各个场景和任务目标。本文一方面提炼出了基于文本数据挖掘的分析架构,另一方面提升了客户投诉反馈工单的数据质量,并对上海通用五菱公司文本数据体系建设和流程环节质量提升有积极的促进作用。As typical unstructured data,text data exists in all aspects of the whole business process of automobile enterprises. The case study was carried out against the background of the above GM Wuling "Voice of Customer" dataset. First,the text data was preprocessed and a professional dictionary in the automotive field was built. Then,the semantic features obtained from the word vector training model were input into the machine learning models in the algorithm pool for prediction and classification,and ultimately served the application scenarios. On the one hand,this paper proposed an automatic label algorithm of text data fed back by car enterprises′ customers. On the other hand,it improved the data quality of customer complaints and feedback,which has a positive role in promoting the construction of Wuling text data system and the quality improvement.

关 键 词：客诉文本 自然语言处理 文本挖掘 质量提升 

分 类 号：C934[经济管理—管理学]