A Model for Quality Control of Customer Service

In E-commerce industry, customer service is with a very high quality, because of comprehensive and strict training of customer service staffs. With statistics on manual sampling results in abundant history logs, just about 0.2% answers of customer service staffs show attitudinal problems. Furthermore, even for those 0.2% answers, they can hardly be detected by some obvious rules. Therefore, present manual sampling method is very inefficient and time-consuming. In this paper we proposed a problem detection model for quality control of customer service. Experimental results on a dataset from a real-world industrial customer service Alime Assist show the model’s effect.