Call center text mining approach

Nowadays, the ability to convert call records from voice to text makes it possible to apply text mining methods to extract information from calls. In this study, it is aimed not only to evaluate the sentiment (positive/negative) of the calls in general, but also to measure the customer satisfaction and representative's performance by using call record texts. New features have been extracted from texts using text mining methods. Using the features extracted, prediction models were developed to evaluate the contents of call records by classification and regression methods. As a result of this study, it is planned to utilize the prediction models developed in Turk Telekom's call centers.