Live agent recommendation for a human-robot symbiosis conversation system

Patent metadata

Field	Value
Publication number	US-11509770-B2
Application number	US-201816141223-A
Country	US
Kind code	B2
Filing date	Sep 25, 2018
Priority date	Sep 25, 2018
Publication date	Nov 22, 2022
Grant date	Nov 22, 2022

Abstract

A computer-implemented method is presented for selecting a preferred live agent from a plurality of live agents. The method includes constructing, via the processor, a human expertise matrix pertaining to each of the plurality of live agents by determining an average net promoter score (NPS) for each of the plurality of live agents for each category of a plurality of categories, and in response to a voice call by a user, determining, via the processor, a predicted human expertise on average by collectively assessing the human expertise matrix, a predicted NPS derived from a first deep neural network, and a predicted category derived from a second deep neural network. The method further includes, based on the predicted human expertise on average determined, triggering communication via the live agent communication network between the user and the preferred live agent to initiate a conversation between the user and the preferred live agent.

First claim

What is claimed is:

1. A computer-implemented method executed on a processor for selecting a preferred live agent from a plurality of live agents linked within a live agent communication network, the computer-implemented method comprising:
employing an artificial intelligence engine to perform operations including:
constructing a human expertise matrix pertaining to each of the plurality of live agents by determining an average net promoter score (NPS) for each of the plurality of live agents for each category of a plurality of categories;
in response to a voice call by a user, implementing a first deep neural network to determine a predicted NPS and implementing a second deep neural network to determine a predicted category;
determining a predicted human expertise on average by collectively assessing the human expertise matrix, the predicted NPS, and the predicted category; and
based on the predicted human expertise on average determined, triggering communication via the live agent communication network between the user and the preferred live agent to initiate a conversation between the user and the preferred live agent.

2. The method of claim 1 , wherein overall service quality in the live agent communication network is maximized by improving a selection process of the preferred live agent, the selection process improved by maximizing the average NPS for each live agent of the plurality of live agents based on good overall agent performance for multiple types of categories.

3. The method of claim 1 , wherein the first deep neural network is employed to compute the predicted NPS and the second deep neural network is employed to compute the predicted category.

4. The method of claim 1 , further comprising computing a loss function for each of the first and second deep neural networks.

5. The method of claim 4 , further comprising updating a neuron weight of each of the first and second deep neural networks.

6. The method of claim 1 , wherein the first and second deep neural networks are employed to create a conversation categorization model.

7. The method of claim 6 , wherein a stochastic prediction of each category for each of the plurality of agents is determined based on the conversation categorization model.

8. A non-transitory computer-readable storage medium comprising a computer-readable program executed on a processor in a data processing system for selecting a preferred live agent from a plurality of live agents, wherein the computer-readable program when executed on the processor causes the data processing system to perform the steps of:
employing an artificial intelligence engine to perform operations including:
constructing a human expertise matrix pertaining to each of the plurality of live agents by determining an average net promoter score (NPS) for each of the plurality of live agents for each category of a plurality of categories;
in response to a voice call by a user, implementing a first deep neural network to determine a predicted NPS and implementing a second deep neural network to determine a predicted category;
determining a predicted human expertise on average by collectively assessing the human expertise matrix, the predicted NPS, and the predicted category; and
based on the predicted human expertise on average determined, triggering communication via the live agent communication network between the user and the preferred live agent to initiate a conversation between the user and the preferred live agent.

9. The non-transitory computer-readable storage medium of claim 8 , wherein overall service quality in the live agent communication network is maximized by improving a selection process of the preferred live agent, the selection process improved by maximizing the average NPS for each live agent of the plurality of live agents based on good overall agent performance for multiple types of categories.

10. The non-transitory computer-readable storage medium of claim 8 , wherein the first deep neural network is employed to compute the predicted NPS and the second deep neural network is employed to compute the predicted category.

11. The non-transitory computer-readable storage medium of claim 8 , wherein a loss function is computed for each of the first and second deep neural networks.

12. The non-transitory computer-readable storage medium of claim 11 , wherein a neuron weight is updated for each of the first and second deep neural networks.

13. The non-transitory computer-readable storage medium of claim 8 , wherein the first and second deep neural networks are employed to create a conversation categorization model.

14. The non-transitory computer-readable storage medium of claim 13 , wherein a stochastic prediction of each category for each of the plurality of agents is determined based on the conversation categorization model.

15. A system for selecting a preferred live agent from a plurality of live agents, the system comprising:
a memory;
one or more processors in communication with the memory; and
an artificial intelligence engine to perform operations including:
constructing a human expertise matrix pertaining to each of the plurality of live agents by determining an average net promoter score (NPS) for each of the plurality of live agents for each category of a plurality of categories;
in response to a voice call by a user, implementing a first deep neural network to determine a predicted NPS and implementing a second deep neural network to determine a predicted category;
determining a predicted human expertise on average by collectively assessing the human expertise matrix, the predicted NPS, and the predicted category; and
based on the predicted human expertise on average determined, triggering communication via the live agent communication network between the user and the preferred live agent to initiate a conversation between the user and the preferred live agent.

16. The system of claim 15 , wherein overall service quality in the live agent communication network is maximized by improving a selection process of the preferred live agent, the selection process improved by maximizing the average NPS for each live agent of the plurality of live agents based on good overall agent performance for multiple types of categories.

17. The system of claim 15 , wherein the first deep neural network is employed to compute the predicted NPS and the second deep neural network is employed to compute the predicted category.

18. The system of claim 15 , wherein a loss function is computed for each of the first and second deep neural networks.

19. The system of claim 18 , wherein a neuron weight is updated for each of the first and second deep neural networks.

20. The system of claim 15 , wherein the first and second deep neural networks are employed to create a conversation categorization model.

Assignees

• IBM

Inventors

• Ma Jie
• Zhou Xin
• Chen Hao
• Mohammed Rachel
• Davis Christopher J
• Kancharla Sharath
• Shu Zhongzheng
• Knoertzer Manon
• Guan Ran

Classifications

• G06N3/045

  Combinations of networks · CPC title

• G06Q10/063112 Primary

  Skill-based matching of a person or a group to a task · CPC title

• H04M2203/40

  related to call centers · CPC title

• G06N3/006

  based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO] · CPC title

• G10L15/16

  using artificial neural networks · CPC title