Systems and methods for a virtual agent in a cloud computing environment

Patent metadata

Field	Value
Publication number	US-12182517-B2
Application number	US-202318462092-A
Country	US
Kind code	B2
Filing date	Sep 6, 2023
Priority date	Mar 19, 2019
Publication date	Dec 31, 2024
Grant date	Dec 31, 2024

Abstract

Present embodiments are directed to a virtual agent with improved natural language understanding (NLU) capabilities. The disclosed virtual agent enables topic selection and topic changes during natural language exchanges with a user. The virtual agent is designed to select suitable topic flows to execute based on intents identified in received user utterances, including selection of an initial topic flow in response to a topic identified in a first user utterance, as well as switching between topic flows mid-conversation based on identified topic changes. The virtual agent is also capable of considering all intents and entities conveyed during the conversation, which enables the virtual agent to avoid prompting the user to provide redundant information. Furthermore, the virtual agent is capable of executing topic flows as part of a global topic flow, which enables the virtual agent to perform a number of predefined activities as part of each interaction with the user.

First claim

The invention claimed is:

1. A method of operating a virtual agent, comprising:
executing a global topic flow that defines a series of nodes having instructions that are executed for each natural language exchange between the virtual agent and a plurality of users, wherein executing the global topic flow comprises:
executing a topic discovery node of the global topic flow, wherein executing the topic discovery node comprises:
receiving a first user utterance;
providing the first user utterance to a natural language understanding (NLU) service;
receiving, from the NLU service, a first scored set of intents and entities derived from the first user utterance;
selecting a first topic flow in response to determining that a first intent from the first scored set of intents and entities is associated with the first topic flow;
applying first entities from the first scored set of intents and entities to first input nodes of the first topic flow;
executing the first topic flow to respond to the first user utterance;
receiving a second user utterance during execution of the first topic flow;
selecting a second topic flow in response to determining that a second intent from a second scored set of intents and entities is associated with the second topic flow, wherein the second scored set of intents and entities is derived from the second user utterance;
applying the first entities from the first scored set of intents and entities, second entities from the second scored set of intents and entities, or a combination thereof, to second input nodes of the second topic flow;
executing the second topic flow to respond to the second user utterance; and
determining whether to resume execution of the first topic flow after completing execution of the second topic flow.

2. The method of claim 1 , wherein executing the second topic flow of the topic discovery node of the global topic flow comprises:
providing the second user utterance to the NLU service; and
receiving, from the NLU service, the second scored set of intents and entities derived from the second user utterance.

3. The method of claim 1 , wherein the topic discovery node of the global topic flow is configured to resume execution of the first topic flow after completing execution of the second topic flow based on a determination step associated with the first topic flow providing an indication to resume execution of the first topic flow.

4. The method of claim 1 , wherein executing the global topic flow comprises executing a greetings node of the global topic flow that outputs a predefined greeting before executing the topic discovery node.

5. The method of claim 1 , wherein executing the global topic flow comprises:
executing an additional topic node of the global topic flow that outputs a predefined prompt regarding one or more additional topics to be discussed after executing the topic discovery node;
receiving user input in response to the predefined prompt; and then
executing a decision node of the global topic flow that repeats execution of the topic discovery node based on the user input received in response to the predefined prompt.

6. The method of claim 5 , wherein the user input is a second third user utterance, and comprising:
providing the third user utterance to the NLU service;
receiving, from the NLU service, a third scored set of intents and entities derived from the third user utterance; and
identifying an additional topicof the one or more additional topics to be discussed based on the third scored set of intents and entities derived from the third user utterance.

7. The method of claim 6 , wherein the predefined prompt includes user interface elements, and wherein receiving the user input comprises receiving the user input based on user interactions with the user interface elements of the predefined prompt.

8. The method of claim 5 , wherein executing the global topic flow comprises:
after executing the decision node, executing a survey node of the global topic flow to output a predefined survey question and to receive and store a second user input in response to the predefined survey question.

9. A virtual agent system, comprising:
at least one memory configured to store a global topic flow that defines a series of nodes having instructions that are executed for each natural language exchange between the virtual agent system and a plurality of users; and
at least one processor configured to execute the global topic flow to cause the virtual agent system to perform actions comprising:
executing a topic discovery node, wherein the topic discovery node is configured to:
receive a first user utterance from a user;
provide the first user utterance to a natural language understanding (NLU) service;
receive, from the NLU service, a first scored set of intents and entities derived from the first user utterance;
select a first topic flow in response to determining that a first intent from the first scored set of intents and entities is associated with the first topic flow;
apply first entities from the first scored set of intents and entities to first input nodes of the first topic flow;
execute the first topic flow to respond to the first user utterance;
receive a second user utterance during execution of the first topic flow;
select a second topic flow in response to determining that a second intent from a second scored set of intents and entities is associated with the second topic flow, wherein the second scored set of intents and entities is derived from the second user utterance;
apply the first entities from the first scored set of intents and entities, second entities from the second scored set of intents and entities, or a combination thereof, to second input nodes of the second topic flow;
execute the second topic flow to response to the second user utterance; and
determine whether to resume execution of the first topic flow after completing execution of the second topic flow.

10. The system of claim 9 , wherein the series of nodes includes a greetings node that is executed before the topic discovery node in the global topic flow, wherein the greetings node of the global topic flow is configured to output a predefined greeting to the user.

11. The system of claim 9 , wherein the series of nodes includes an additional topics node that is executed after the topic discovery node in the global topic flow, wherein the additional topics node of the global topic flow is configured to:
output a predefined prompt to the user asking whether the user has additional topics to be discussed; and
receive input from the user in response to the predefined prompt.

12. The system of claim 11 , wherein the series of nodes includes a decision node that is executed after the additional topics node in the global topic flow, wherein the decision node of the global topic flow is configured to repeat execution of the topic discovery node based on the input received from the user in response to the predefined prompt.

13. The system of claim 12 , wherein the series of nodes includes a survey node that is executed after the decision node in the global topic flow, wherein the survey node is configured to:
output a predefined survey question to the user; and
receive and store user input in response to the predefined survey question.

14. A non-transitory, computer-readable medium storing instructions executable by a processor of a computing system, the instructions comprising instructions to:
execute a global topic flow that defines a series of nodes having instructions that are executed for each natural language exchange between a virtual agent a

Assignees

• Servicenow Inc

Inventors

• Asba Nabil Abu
• Griffin Mark David
• Jain Anuj

Classifications

• G06F40/279

  Recognition of textual entities · CPC title

• G10L2015/223

  Execution procedure of a spoken command · CPC title

• G10L15/22

  Procedures used during a speech recognition process, e.g. man-machine dialogue · CPC title

• G10L15/1822

  Parsing for meaning understanding · CPC title

• G06F40/35 Primary

  Discourse or dialogue representation · CPC title