Systems and methods for dynamically expanding natural language processing agent capacity

• US 10,839,167 B2
• Filed: 12/04/2018
• Issued: 11/17/2020
• Est. Priority Date: 12/04/2018
• Status: Active Grant

• Current Assignee: Verizon Patent and Licensing Incorporated (Verizon Communications Inc.)
• Original Assignee: Verizon Patent and Licensing Incorporated (Verizon Communications Inc.)
• Inventors: Mahajan, Trisha, Soltys, Stephen, Razzano, Neil Thomas, Shanmugam, Sankar
• Primary Examiner(s): Gauthier, Gerald
• Application Number:
  US16/209,783
• Publication Number:
  US 20200175118A1
• Time in Patent Office: 714 Days
• Field of Search: 704 9, 704257, 704258, 704275, 705 5, 707776, 718102, 37926507, 706 11
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30
• CPC Class Codes: G10L 15/1822 Parsing for meaning understanding
• CPC Class Codes: G10L 15/183 using context dependencies, e.g. language models
• CPC Class Codes: G10L 17/00 Speaker identification or verification techniques
• CPC Class Codes: G10L 17/22 Interactive procedures; Man-machine interfaces

Abstract

A system described herein may provide for the adaptation and/or expansion of a natural language processing (“NLP”) platform, that supports only a limited quantity of intents, such that the described system may support an unlimited (or nearly unlimited) quantity of intents. For example, a hierarchical structure of agents may be used, where each agent includes multiple intents. A top-level (e.g., master) agent may handle initial user interactions, and may indicate a next-level agent to handle subsequent interactions.

Claims

1. A device, comprising:

a non-transitory computer-readable medium storing a set of processor-executable instructions; and

one or more processors configured to execute the set of processor-executable instructions, wherein executing the set of processor-executable instructions causes the one or more processors to;

receive first user input associated with a particular user;

determine, based on receiving the first user input, that a communication session is not active for the particular user;

provide, based on the determination that a communication session is not active for the particular user, the first user input to a first agent that includes a first set of intents, wherein at least a first intent of the first set of intents specifies;

one or more conditions to compare to the first user input, wherein a matching of the one or more conditions to the first user input indicates that the first intent should be selected from the first set of intents to handle the first user input,

one or more actions to perform when the first intent is selected, and

an identification of a second agent that is different from the first agent, wherein the second agent is a particular agent from a set of candidate agents, wherein each agent of the set of candidate agents is associated with a different respective set of intents;

receive, from the first agent, the identification of the second agent to handle subsequent user input for the particular user;

receive second user input associated with the particular user, after receiving the first user input;

determine, based on receiving the second user input, that the communication session is active for the particular user; and

provide, based on the determination that the communication session is active for the particular user, the second user input to the second agent, wherein the second agent is associated with a second set of intents that is different from the first set of intents and includes at least a second intent that includes at least a portion of the second user input as a condition for selection of the second intent to handle the second user input.

2. The device of claim 1, wherein executing the processor-executable instructions further causes the one or more processors to:

receive one or more context parameters, for the particular user, that have been provided by the selected first intent;

identify the one or more context parameters based on receiving the second user input and identifying that the communication session is active for the particular user; and

further provide the one or more context parameters to the second agent, wherein the second intent further includes at least a portion of the one or more context parameters as the condition for selection of the second intent.

3. The device of claim 1, wherein executing the processor-executable instructions further causes the one or more processors to:

store, based on receiving the identification of the second agent from the first agent, mapping information that correlates an identifier of the particular user to the second agent; and

select, based on the mapping information, the second agent, from a plurality of agents, to handle the second user input, wherein the providing of the second user input to the second agent is based on the selection of the second agent.

4. The device of claim 1, wherein the one or more actions includes providing a response to a user device associated with the particular user.

5. The device of claim 1, wherein the first and second agents are provided via a natural language processing platform that enforces a limit on a quantity of intents included in each of the first and second agents.

6. The device of claim 1, wherein executing the processor-executable instructions further causes the one or more processors to:

maintain separate sets of context parameters for separate users, whereby the user is associated with a first set of context parameters and wherein another user is associated with a second set of context parameters.

7. The device of claim 1, wherein the processor-executable instructions, to determine whether a communication session is active for the particular user, include processor-executable instructions to:

determine whether the device stores a set of context parameters for the particular user.

8. A non-transitory computer-readable medium, storing a set of processor-executable instructions, which, when executed by one or more processors, cause the one or more processors to:

receive first user input associated with a particular user;

determine, based on receiving the first user input, that a communication session is not active for the particular user;

provide, based on the determination that a communication session is not active for the particular user, the first user input to a first agent that includes a first set of intents, wherein at least a first intent of the first set of intents specifies;

one or more conditions to compare to the first user input, wherein a matching of the one or more conditions to the first user input indicates that the first intent should be selected from the first set of intents to handle the first user input,

one or more actions to perform when the first intent is selected, and

an identification of a second agent that is different from the first agent, wherein the second agent is a particular agent from a set of candidate agents, wherein each agent of the set of candidate agents is associated with a different respective set of intents;

receive, from the first agent, the identification of the second agent to handle subsequent user input for the particular user;

receive second user input associated with the particular user, after receiving the first user input;

determine, based on receiving the second user input, that the communication session is active for the particular user; and

provide, based on the determination that the communication session is active for the particular user, the second user input to the second agent, wherein the second agent is associated with a second set of intents that is different from the first set of intents and includes at least a second intent that includes at least a portion of the second user input as a condition for selection of the second intent to handle the second user input.

9. The non-transitory computer-readable medium of claim 8, wherein the processor-executable instructions further include processor-executable instructions to:

receive one or more context parameters, for the particular user, that have been provided by the selected first intent;

identify the one or more context parameters based on receiving the second user input and identifying that the communication session is active for the particular user; and

further provide the one or more context parameters to the second agent, wherein the second intent further includes at least a portion of the one or more context parameters as the condition for selection of the second intent.

10. The non-transitory computer-readable medium of claim 8, wherein the processor-executable instructions further include processor-executable instructions to:

store, based on receiving the identification of the second agent from the first agent, mapping information that correlates an identifier of the particular user to the second agent; and

select, based on the mapping information, the second agent, from a plurality of agents, to handle the second user input, wherein the providing of the second user input to the second agent is based on the selection of the second agent.

11. The non-transitory computer-readable medium of claim 8, wherein the one or more actions includes providing a response to a user device associated with the particular user.

12. The non-transitory computer-readable medium of claim 8, wherein the first and second agents are provided via a natural language processing platform that enforces a limit on a quantity of intents included in each of the first and second agents.

13. The non-transitory computer-readable medium of claim 8, wherein the processor-executable instructions further include processor-executable instructions to:

maintain separate sets of context parameters for separate users, whereby the user is associated with a first set of context parameters and wherein another user is associated with a second set of context parameters.

14. A method, comprising:

receiving, by a device, first user input associated with a particular user;

determining, by the device and based on receiving the first user input, that a communication session is not active for the particular user;

providing, by the device, based on the determination that a communication session is not active for the particular user, the first user input to a first agent that includes a first set of intents, wherein at least a first intent of the first set of intents specifies;

one or more conditions to compare to the first user input, wherein a matching of the one or more conditions to the first user input indicates that the first intent should be selected from the first set of intents to handle the first user input,

one or more actions to perform when the first intent is selected, and

an identification of a second agent that is different from the first agent, wherein the second agent is a particular agent from a set of candidate agents, wherein each agent of the set of candidate agents is associated with a different respective set of intents;

receiving, by the device and from the first agent, the identification of the second agent to handle subsequent user input for the particular user;

receiving, by the device, second user input associated with the particular user, after receiving the first user input;

determining, by the device and based on receiving the second user input, that the communication session is active for the particular user; and

providing, by the device and based on the determination that the communication session is active for the particular user, the second user input to the second agent, wherein the second agent is associated with a second set of intents that is different from the first set of intents and includes at least a second intent that includes at least a portion of the second user input as a condition for selection of the second intent to handle the second user input.

15. The method of claim 14, further comprising:

receiving one or more context parameters, for the particular user, that have been provided by the selected first intent;

identifying the one or more context parameters based on receiving the second user input and identifying that the communication session is active for the particular user; and

further providing the one or more context parameters to the second agent, wherein the second intent further includes at least a portion of the one or more context parameters as the condition for selection of the second intent.

16. The method of claim 14, further comprising:

store, based on receiving the identification of the second agent from the first agent, mapping information that correlates an identifier of the particular user to the second agent; and

select, based on the mapping information, the second agent, from a plurality of agents, to handle the second user input, wherein the providing of the second user input to the second agent is based on the selection of the second agent.

17. The method of claim 14, wherein the one or more actions includes providing a response to a user device associated with the particular user.

18. The method of claim 14, wherein the first and second agents are provided via a natural language processing platform that enforces a limit on a quantity of intents included in each of the first and second agents.

19. The method of claim 14, further comprising:

maintaining separate sets of context parameters for separate users, whereby the user is associated with a first set of context parameters and wherein another user is associated with a second set of context parameters.

20. The method of claim 14, wherein determining whether a communication session is active for the particular user, includes:

determining whether the device stores a set of context parameters for the particular user.