Context-based natural language processing

• US 10,482,184 B2
• Filed: 03/08/2016
• Issued: 11/19/2019
• Est. Priority Date: 03/08/2015
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google LLC (Alphabet Inc.)
• Inventors: Gelfenbeyn, Ilya Gennadyevich, Goncharuk, Artem, Sirotin, Pavel Aleksandrovich
• Primary Examiner(s): Serrou, Abdelali
• Application Number:
  US15/063,757
• Publication Number:
  US 20160259775A1
• Time in Patent Office: 1,351 Days
• CPC Class Codes: G06F 40/279 Recognition of textual entities
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence
• CPC Class Codes: G10L 2015/228 of application context

Abstract

A method for context-based natural language processing is disclosed herein. The method comprises maintaining a plurality of dialog system rules, receiving a user request from a Dialog System Interface, receiving one or more attributes associated with the user request from the Dialog System Interface or a user device, and identifying a type of context associated with the user request based on the user request and the one or more attributes. A context label is assigned to the user request associated with the type of context. Based on the context label and the user request, a particular dialog system rule is selected from the plurality of dialog system rules. A response to the user request is generated by applying the dialog system rule to at least a part of the user request.

Claims

1. A method for context-based natural language processing, comprising:

maintaining, by a computer comprising at least one processor and a memory storing processor-executable codes, a plurality of dialog system rules;

receiving, by the computer, a user request from a Dialog System Interface, wherein the user request is based on input provided by a user during a current user session via the Dialog System Interface, and wherein the user request includes speech input;

determining, by the computer, user request text for the user request, wherein determining the user request text includes performing speech recognition to convert the speech input to the user request text;

receiving, by the computer, one or more attributes associated with the user request from the Dialog System Interface or a user device, wherein the one or more attributes associated with the user request include at least one prior context label based on a preceding user request from the Dialog System Interface, and wherein the preceding user request, on which the at least one prior context label is based, is based on prior spoken input provided by the user during the current user session via the Dialog System Interface;

identifying, by the computer, a type of context associated with at least a portion of the user request text converted from the speech input, wherein identifying the type of context is based on the user request and the one or more attributes;

assigning, by the computer, a context label to the user request text converted from the speech input, the context label being associated with the type of context;

selecting, by the computer, from the plurality of dialog system rules, a dialog system rule which is associated with the context label and the user request text converted from the speech input; and

generating, by the computer, a response to the user request by applying the dialog system rule to at least a part of the user request text converted from the speech input.

2. The method of claim 1, further comprising delivering, through the Dialog System Interface, the response to the user request.

3. The method of claim 1, wherein the response includes one or more of:

performing an actionable command, opening a webpage associated with the user request, scheduling an event in a calendar mobile application, writing and sending an email, searching for content at a web search service, building a route in a navigational software application, generating an audio message, generating a video, providing a text message, and providing a clarifying question, the clarifying question being generated based on the dialog system rule.

4. The method of claim 1, further comprising acquiring additional data to assign the context label, wherein the additional data is associated with the user request from one or more of the following:

the user device, a web resource, a web service, and a database.

5. The method of claim 1, further comprising pre-processing the user request by the Dialog System Interface.

6. The method of claim 1, wherein the one or more attributes further include one or more of:

a geographical location of the user, a date, a time, a software application running on the user device, a type of the user device, a make of the user device, weather conditions in the geographical location of the user, a motion of the user, a direction of movement of the user, data contained on the user device, one or more scheduled events of the user, one or more errands of the user, a physical condition of the user, and a health condition of the user.

7. The method of claim 1, further comprising:

receiving, from the user, a command to force expiration of the context label; and

based on the command, discarding the context label.

8. The method of claim 1, further comprising:

tracking one or more previous context labels for one or more previous user requests and one or more previous dialog system rules for the one or more previous user requests; and

based on the tracking, assigning a further context label to the user request text converted from the speech input.

9. The method of claim 8, further comprising:

aggregating the context label and the one or more previous context labels in a global user context;

prioritizing the global user context over the context label and one or more previous context labels; and

processing the user request text converted from the speech input using the global user context.

10. The method of claim 1, further comprising:

determining that the context label is associated with a predicate, wherein the predicate implements arbitrary logic on the type of the context and context variables; and

based on the determination, applying an arbitrary logic to the user request.

11. The method of claim 1, further comprising:

determining, by the computer, that the user request cannot be understood without the dialog system rule that is associated with the context label;

wherein generating the response to the user request by applying the dialog system rule to at least a part of the user request text converted from the speech input is based on determining that the user request cannot be understood without the dialog system rule that is associated with the context label.

12. The method of claim 1, wherein selecting the dialog system rule which is associated with the context label and the user request text converted from the speech input comprises:

determining an intent of the user based on the user request text converted from the speech input of the user request;

determining the intent of the user matches an intent of the dialog system rule, which is associated with the context label and the user request text converted from the speech input; and

selecting, based on determining the intent of the user matches the intent of the dialog system rule, the dialog system rule which is associated with the context label and the user request text converted from the speech input.

13. A system for context-based natural language processing, comprising:

a computer including at least one processor and a memory storing processor-executable codes, the computer configured to;

maintain a plurality of dialog system rules;

receive a user request from a Dialog System Interface wherein the user request is based on input provided by a user during a current user session via the Dialog System Interface, and wherein the user request includes speech input;

determine a user request text for the user request, wherein determining the user request text includes performing speech recognition to convert the speech input to the user request text;

receive one or more attributes associated with the user request from the Dialog System Interface or a user device, wherein the one or more attributes associated with the user request include at least one prior context label based on a preceding user request from the Dialog System Interface, and wherein the preceding user request, on which the at least one prior context label is based, is based on prior spoken input provided by the user during the current user session via the Dialog System Interface;

identify a type of context associated with at least a portion of the user request text converted from the speech input, wherein identifying the type of context is based on the user request and the one or more attributes;

assign a context label to the user request text converted from the speech input, the context label being associated with the type of context;

select, from the plurality of dialog system rules, a dialog system rule which is associated with the context label and the user request text converted from the speech input; and

generate a response to the user request by applying the dialog system rule to at least a part of the user request text converted from the speech input.

14. The system of claim 13, further comprising a context database, wherein the context database is automatically populated with additional terms, keywords, phrases, variables, and expressions using one or more machine-learning techniques based on the user request text converted from the speech input.

15. The system of claim 13, wherein the user request is provided as a natural language speech, the natural language speech being transformed to the user request text converted from the speech input.

16. The system of claim 13, wherein the context label is associated with one or more of the following:

a plurality of context-defining terms, a plurality of keywords, a plurality of phrases, a plurality of variables, and a plurality of expressions.

17. The system of claim 13, wherein the context label is set to a Boolean expression.

18. The system of claim 13, wherein the context label expires after a predetermined period since assigning the context label to the user request text converted from the speech input.

19. The system of claim 13, wherein the context label expires after a dynamically varied period based on one or more circumstances.

20. A non-transitory processor-readable medium having instructions stored thereon, which when executed by one or more processors, cause the one or more processors to implement a method, comprising:

maintaining a plurality of dialog system rules;

receiving a user request from a Dialog System Interface, wherein the user request is based on input provided by a user during a current user session via the Dialog System Interface, and wherein the user request includes speech input;

determining a user request text for the user request, wherein determining the user request text includes performing speech recognition to convert the speech input to the user request text;

receiving one or more attributes associated with the user request from the Dialog System Interface or a user device, wherein the one or more attributes associated with the user request include at least one prior context label based on a preceding user request from the Dialog System Interface, and wherein the preceding user request, on which the at least one prior context label is based, is based on prior spoken input provided by the user during the current user session via the Dialog System Interface;

identifying a type of context associated with at least a portion of the user request text converted from the speech input, wherein identifying the type of context is based on the user request text and the one or more attributes;

assigning a context label to the user request text converted from the speech input, the context label being associated with the type of context;

selecting, from the plurality of dialog rules, a dialog system rule which is associated with the context label and the user request text converted from the speech input; and

generating a response to the user request by applying the dialog system rule to at least a part of the user request text converted from the speech input.