A method and device for processing recorded data of customer service calls

Abstract

The invention provides a customer service call recording data processing method and device. Customer service call recording data record conversations between clients and customer service representatives. The method comprises, according to recording data to be evaluated, acquiring the conversation rhythm features of a client and a customer service representative in the recording data to be evaluated; according to the conversation rhythm features of the client and the customer service representative in the recording data to be evaluated and through a pre-determined satisfaction evaluation model, determining a satisfaction evaluation result of the recording data to be evaluated. The customer service call recording data processing method can automatically evaluate the users' satisfaction of customer service call recordings on a conversation rhythm level, widen the feature space of satisfaction evaluation, and better provide technical support for customer service hotline management and operation from a new perspective.

Description

technical field [0001] The invention relates to the field of communication technology, in particular to a method and device for processing recorded data of customer service calls. Background technique [0002] Customer service hotline recording satisfaction evaluation was mainly obtained directly from the user side through Interactive Voice Response (IVR) or sending short messages. However, due to the low user participation and response rate, most hotline calls Satisfaction evaluation was not completed. Therefore, automatic satisfaction evaluation is produced, but the current automatic satisfaction evaluation mainly converts the recording data into recognized text through the speech recognition engine, and then models the related situation of text emotion words to complete the automatic satisfaction evaluation of the recording . Among them, speech analysis technology and machine learning technology are mainly involved. [0003] Speech analysis technology decomposes the in...

AI Technical Summary

Problems solved by technology

[0009] However, in the above-mentioned three schemes, scheme one uses the speech recognition engine to obtain the recognized text, and processes the text information to obtain the user satisfaction information. In fact, the speech satisfaction information in the original recording is lost, and many users are not satisfied. Not fully reflected, affecting the accuracy of satisfaction evaluation

Scenario 2 Judging the user’s satisfaction simply from voice emotion does not give a particularly good evaluation effect. On the one hand, emotion itself is a vague and subjective concept, and the evaluation criteria vary from person to person; Contains semantic information, so it is necessary to use joint features to better complete automatic evaluation tasks

[0010] However, although the automatic satisfaction evaluation has been greatly improved by combining the speech emotional features and the recognized text emotional features in the third scheme, since emotion and satisfaction itself are a vague subjective concept, the current effect is still relatively large. room for improvement, it is necessary to combine more types of features to improve the existing automatic evaluation effect

Images