Context carryover across tasks for assistant systems

Patent metadata

Field	Value
Publication number	US-12299755-B2
Application number	US-202318462101-A
Country	US
Kind code	B2
Filing date	Sep 6, 2023
Priority date	Oct 18, 2019
Publication date	May 13, 2025
Grant date	May 13, 2025

Abstract

In one embodiment, a method includes maintaining contextual information from a first dialog session, wherein the contextual information comprises identifiers of entities and slots, receiving a user request during the first dialog session, determining a context carryover is required for an agent to execute a task associated with the user request, determining a first subset of the identifiers of the entities and a second subset of the slots of the contextual information from the first dialog session are shareable with the agent based on the agent and first resources of the identifiers of the entities and second resources of the slots, and executing the task by the agent based on the first subset of the identifiers of the entities and the second subset of the slots of the shareable contextual information.

First claim

What is claimed is:

1. A method comprising, by a client system:
maintaining, by a context engine, contextual information from a first dialog session, wherein the contextual information comprises one or more identifiers of one or more entities, respectively, and one or more slots;
receiving, at the client system, a user request during the first dialog session;
determining a context carryover is required for an agent to execute a task associated with the user request;
determining, based on the agent and first resources of the one or more identifiers of the one or more entities and second resources of the one or more slots, a first subset of the one or more identifiers of the one or more entities and a second subset of the one or more slots of the contextual information from the first dialog session are shareable with the agent; and
executing, by the agent, the task based on the first subset of the one or more identifiers of the one or more entities and the second subset of the one or more slots of the shareable contextual information.

2. The method of claim 1 , further comprising:
receiving, at the client system, a prior user request associated with a prior task;
executing the prior task by a prior agent;
presenting, at the client system, a prior response generated based on the execution of the prior task; and
storing the contextual information from the prior user request in the context engine.

3. The method of claim 2 , wherein the prior agent and the agent are different agents.

4. The method of claim 2 , wherein the prior agent is associated with a first-party system and the agent is associated with a third-party system that is separate from and external to the first-party system.

5. The method of claim 2 , wherein the user request comprises an ambiguous mention, and wherein the method further comprises:
determining the ambiguous mention corresponds to one or more data items associated with the contextual information.

6. The method of claim 5 , further comprising providing the one or more data items and the first subset of the one or more identifiers of the one or more entities and the second subset of the one or more slots to the agent, wherein the providing comprises:
accessing the one or more data items and the first subset of the one or more identifiers of the one or more entities and the second subset of the one or more slots by the prior agent; and
sending the one or more data items and the first subset of the one or more identifiers of the one or more entities and the second subset of the one or more slots from the prior agent to the agent.

7. The method of claim 6 , further comprising:
determining whether the context carryover is permitted per one or more privacy settings, wherein the one or more data items are provided to the agent responsive to determining that the context carryover is permitted per the one or more privacy settings.

8. The method of claim 2 , wherein the prior task and the task are associated with different domains.

9. The method of claim 2 , wherein the prior user request and the user request are associated with different modalities.

10. The method of claim 9 , wherein determining the context carryover is required is based on a difference between the modalities associated with the prior user request and the user request.

11. The method of claim 1 , further comprising:
determining one or more intents and one or more slots associated with the user request.

12. The method of claim 11 , wherein determining the context carryover is required is based on one or more of an intent-confidence score associated with each intent or a slot-confidence score associated with each slot.

13. The method of claim 1 , wherein determining the context carryover is required is based on one or more task policies associated with the task.

14. The method of claim 1 , wherein determining the context carryover is required is based on one or more machine-learning models.

15. The method of claim 14 , wherein the machine-learning models are trained based on training data generated by a stitching multi-task conversation model.

16. The method of claim 1 , further comprising:
determining whether a user confirmation is required for the context carryover; and
presenting a prompt to confirm the context carryover responsive to determining that user confirmation is required.

17. The method of claim 1 , wherein the contextual information is stored in a dialog state associated with the user request.

18. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:
maintain, by a context engine, contextual information from a first dialog session, wherein the contextual information comprises one or more identifiers of one or more entities, respectively, and one or more slots;
receive, at a client system, a user request during the first dialog session;
determine a context carryover is required for an agent to execute a task associated with the user request;
determine, based on the agent and first resources of the one or more identifiers of the one or more entities and second resources of the one or more slots, a first subset of the one or more identifiers of the one or more entities and a second subset of the one or more slots of the contextual information from the first dialog session are shareable with the agent; and
execute, by the agent, the task based on the first subset of the one or more identifiers of the one or more entities and the second subset of the one or more slots of the shareable contextual information.

19. A system comprising: one or more processors; and a non-transitory memory coupled to the processors comprising instructions executable by the processors, the processors operable when executing the instructions to:
maintain, by a context engine, contextual information from a first dialog session, wherein the contextual information comprises one or more identifiers of one or more entities, respectively, and one or more slots;
receive, at a client system, a user request during the first dialog session;
determine a context carryover is required for an agent to execute a task associated with the user request;
determine, based on the agent and first resources of the one or more identifiers of the one or more entities and second resources of the one or more slots, a first subset of the one or more identifiers of the one or more entities and a second subset of the one or more slots of the contextual information from the first dialog session are shareable with the agent; and
execute, by the agent, the task based on the first subset of the one or more identifiers of the one or more entities and the second subset of the one or more slots of the shareable contextual information.

Assignees

• Meta Platforms Tech Llc

Inventors

• Liu Xiaohu
• Weng Kai Jun
• Wang Zhiguang
• Crook Paul Anthony

Classifications

• G06Q10/40 Primary

  Business processes related to social networking or social networking services · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/098

  Distributed learning, e.g. federated learning · CPC title

• G06Q10/1093

  Calendar-based scheduling for persons or groups · CPC title

• G10L2015/0631

  Creating reference templates; Clustering · CPC title