Scenario Building, Quality Control

Automation of building scenarios dialogue operators contact center with the company’s customers and further automation of the evaluation of the quality of activity operators contact center requires categorization of a set of dialogues operators and customers, highlight the main themes, the recognition sequences in the dialogue with operators by subscribers. It is planned to use text segmentation into semantically related pieces using topic modeling models.

Input Data

  • Dialogues of the contact center operators with the client
  • Information about the success of the dialogue
  • Priori expert knowledge about topics
  • Assessor markup

Main Results

Topics covered: 41 topics
Topics Quality: 75% accuracy on average

Output Data

  • A set of topics for operator dialogues and customers
  • Count transition between topics for successful and unsuccessful dialogues
  • Topic segmentation tool for dialogue replicas

Integration of business intelligence Bank’s assessment of the quality operators
Increase conversion and share successful dialogs

Examples of topics in the collection:

Willing to talk
Repeated call
Purpose of the call
Credit limit
Interest free period
Bonus program
General issues
Card decision
Card use

Topic graph and the implementation of the example dialogue:

Example markup dialogue in the collection:

О: Alexey Vladimirovich, is it convenient for you to talk at the moment?
А: Yes, yes, speak.
О: Indeed, сour bank employee tried to call you, but the dialogue you have not got. I also want to inform you that we have an offer for a credit card with favorable conditions: a high initial credit limit, an interest-free grace period of up to 50 days and an additional bonus program. Could you be interested in such a proposal?
А: Tell me more about the conditions for obtaining the card.