Customer voice sensor: A comprehensive opinion mining system for call center conversation

Call center is an important intermediary between enterprise and customers. It not only helps customers to solve the problems they are faced with but also allows the enterprise to deeply analyze the customer's voice and make a distinct market positioning. Nowadays customer satisfaction in call center have been attached much importance and studied extensively. However, few researches are actually about analyzing and understanding the data from the mining standpoint. In this paper, a comprehensive opinion mining system of the call center conversation named the customer voice sensor (CVS) is designed and implemented. The proposed system incorporates sentiment classification, information extraction and domain knowledge base techniques. It is able to find out the sentiments of customers and services as well as the intention of the caller. Experimental results on a dataset collected by China Mobile Communication Corporation prove the effectiveness of these techniques.