Templated rule-based data augmentation for intent extraction

Patent metadata

Field	Value
Publication number	US-10970487-B2
Application number	US-201916239218-A
Country	US
Kind code	B2
Filing date	Jan 3, 2019
Priority date	Mar 23, 2018
Publication date	Apr 6, 2021
Grant date	Apr 6, 2021

Abstract

An agent automation system includes a memory configured to store a natural language understanding (NLU) framework and a model, wherein the model includes at least one original meaning representation. The system includes a processor configured to execute instructions of the NLU framework to cause the agent automation system to perform actions including: performing rule-based generalization of the model to generate at least one generalized meaning representation of the model from the at least one original meaning representation of the model; performing rule-based refinement of the model to prune or modify the at least one generalized meaning representation of the model, or the at least one original meaning representation of the model, or a combination thereof; and after performing the rule-based generalization and the rule-based refinement of the model, using the model to extract intents/entities from a received user utterance.

First claim

What is claimed is:

1. An agent automation system, comprising:
a memory configured to store a natural language understanding (NLU) framework, an intent/entity model, and an understanding model, wherein the understanding model comprises a plurality of original meaning representations generated from sample utterances of the intent/entity model; and
a processor configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
performing rule-based generalization of the understanding model to generate at least one generalized meaning representation of the understanding model from the plurality of original meaning representations;
performing rule-based refinement of the understanding model to prune or modify the at least one generalized meaning representation of the understanding model, or the plurality of original meaning representations of the understanding model, or a combination thereof; and
after performing the rule-based generalization and the rule-based refinement of the understanding model:
generating at least one meaning representation for a received user utterance; and
searching the at least one meaning representation of the received user utterance within the understanding model to extract intents/entities from a received user utterance.

2. The system of claim 1 , wherein the memory is configured to store a model augmentation template, wherein the model augmentation template comprises a generalizing rule-set and a refining rule-set, as well as model applicability criteria corresponding to the generalizing rule-set and to the refining rule-set.

3. The system of claim 2 , wherein, to perform rule-based generalization of the model, the processor is configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
selectively applying each generalizing rule of the generalizing rule-set to each portion of the plurality of original meaning representations based on the model applicability criteria corresponding to the generalizing rule-set.

4. The system of claim 2 , wherein, to perform rule-based refinement of the model, the processor is configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
selectively applying each substitution rule of the refining rule-set to each portion of the plurality of original meaning representations and the at least one generalized meaning representation based on the model applicability criteria corresponding to the refining rule-set.

5. The system of claim 2 , wherein, to perform rule-based refinement of the model, the processor is configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
selectively applying each pruning rule of the refining rule-set to remove at least one of the plurality of original meaning representations or the at least one generalized meaning representation of the understanding model based on the model applicability criteria corresponding to the refining rule-set.

6. The system of claim 1 , wherein the intent/entity model comprises defined intents and entities, and comprises sample utterances that are associated with the defined intents and entities.

7. A method of operating an agent automation system, comprising:
generating an understanding model that includes meaning representations for sample utterances of an intent/entity model;
generalizing the understanding model by expanding the meaning representations of the understanding model based on a generalizing rule-set;
refining the understanding model by focusing the meaning representations of the understanding model based on a refining rule-set;
searching the understanding model for a match to at least one meaning representation of a received user utterance to extract intents/entities from the received user utterance; and
performing one or more actions in response to the intents/entities.

8. The method of claim 7 , comprising:
before searching the understanding model or performing the one or more actions:
generating an utterance meaning model from the received user utterance, wherein the utterance meaning model includes the at least one meaning representation of the received user utterance;
generalizing the utterance meaning model by expanding the at least one meaning representation of the utterance meaning model based on the generalizing rule-set; and
refining the utterance meaning model by focusing the at least one meaning representation of the utterance meaning model based on the refining rule-set.

9. The method of claim 7 , wherein expanding the meaning representations of the understanding model comprises:
conditionally applying the generalizing rule-set to the meaning representations of the understanding model to generate additional meaning representations from the meaning representations of the understanding model.

10. The method of claim 7 , wherein focusing the meaning representations of the understanding model comprises:
conditionally applying the refining rule-set to the meaning representations of the understanding model based on model applicability criteria to remove or modify at least a portion of the meaning representations of the understanding model.

11. An agent automation system, comprising:
a memory configured to store a natural language understanding (NLU) framework, a model augmentation template, and an understanding model, wherein the understanding model includes meaning representations for sample utterances of an intent/entity model; and
a processor configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
generalizing the understanding model by selectively expanding the meaning representations of the understanding model based on a generalizing rule-set of the model augmentation template;
refining the understanding model by selectively modifying one or more of the meaning representations of the understanding model based on a refining rule-set of the model augmentation template; and
searching the meaning representations of the understanding model to identify intents/entities from a received user utterance.

12. The system of claim 11 , wherein, to selectively expand the meaning representations of the understanding model, the processor is configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
generating additional meaning representations from the meaning representations of the understanding model based on a grammar-based generalizing rule of the generalizing rule-set.

13. The system of claim 11 , wherein, to selectively modify at least one of the meaning representations of the understanding model, the processor is configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
modifying a structure or a vector of the one or more of the meaning representations of the model based on a substitution rule of the refining rule-set.

14. The system of claim 11 , wherein, to selectively modify at least one of the meaning representations of the understanding model, the processor is configured to execute instructions of the NLU framework to cause the agent automation system to perform actions comprising:
removing the one or more of the meaning representations of the understanding model based on a pruning rule of the refining rule-set.

15. The system of claim 11 , wherein the memory is configured to store an utterance meaning model

Assignees

• Servicenow Inc

Inventors

• Sapugay Edwin
• Madamala Anil Kumar
• Naboka Maxim
• Sunkara Srinivas Satyasai
• Santos Lewis Savio Landry
• Subbarao Murali B

Classifications

• G06F40/30 Primary

  Semantic analysis · CPC title

• G10L15/1807

  using prosody or stress · CPC title

• G06N3/006

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

• G06N5/025

  Extracting rules from data · CPC title

• G06N20/00

  Machine learning · CPC title