The Characteristics Of Question Answering Behavior On The Intelligent Customer Service System

In recent years,artificial intelligence technology has been widely concerned by the whole society because of its high simulation of human consciousness,thinking and human intelligence.The intelligent customer service system is the product of artificial intelligence technology.Intelligent customer service system provides users with online consulting service anytime and anywhere through the intelligent question answering technology,which is widely loved by users and enterprises.Today users are putting higher requirements for the service quality of intelligent customer service platform.In this study,we obtained the intelligent question answering data from October22,2017 to January 24,2018 under the medical insurance category of Tencent Weibao online intelligent customer service platform.After data cleaning,filtering and text segmentation,we first used artificial coding to encode the topic of 5000 randomly selected questions to obtain the preliminary distribution characteristics of user information needs.After that,LDA model is used to automatically extract all the topics recorded in the response,and then the classification and distribution characteristics of the user’s information needs in the topic are obtained by combining the results of the two topic analysis.Secondly,through a series of tag words and their location of,the user’s needs can be judged as satisfied or not in a session.And then this study analyzed the distribution of user’s need satisfaction under different information topics.Third,from the perspective of system,this research analyzed the language characteristics of answers which meet the user’s need,and compared the language characteristics of machine answers and artificial answers;fourth,the study explored the characteristics of question answering modes on the intelligent customer service platform,including the user’s question mode and system’s answer mode,and compared the differences between human and intelligent answer modes.The results illustrated that,in terms of information demand,the current consultation needs of users for medical insurance are mainly focused on the purchase of insurance and insurance compensation for illness after purchase.Secondly,on the whole,the intelligent customer service usually can satisfy the user’s information need,but the satisfaction on the non-medical insurance product issues is slightly worse.In terms of linguistic style,the human answers are detailed and rich,with short sentence length and simple structure,which are easier for users to understand.In contrast,the content of intelligent answers is often longer and hard for users to understand.Finally,users usually only ask one question in a consultation on the customer service platform,and the corresponding human customer service will often give multiple answer messages,while the intelligent answer is usually one message.