Semantic re-ranking of NLU results in conversational dialogue applications

• US 10,540,965 B2
• Filed: 09/11/2017
• Issued: 01/21/2020
• Est. Priority Date: 03/11/2013
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Gandrabur, Simona, Lavallee, Jean-Francois, Tremblay, Real
• Primary Examiner(s): Hang, Vu B
• Application Number:
  US15/700,438
• Publication Number:
  US 20180075846A1
• Time in Patent Office: 862 Days
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G10L 15/1815 Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue

Abstract

Multiple natural language understanding (NLU) interpretation selection models may be generated. The NLU interpretation selection models may include a generic NLU interpretation selection model that is not specialized for a specific set of NLU interpretations type and one or more specialized NLU interpretation selection models, each of which may be specific to a particular set of NLU interpretations type. The specialized NLU interpretation selection model(s) may be utilized to process natural language input data comprising data corresponding to their respective sets of NLU interpretations type(s). The generic NLU interpretation selection model may be utilized to process natural language input data comprising data that does not correspond to the sets of NLU interpretations type(s) associated with the specialized NLU interpretation selection model(s).

Claims

1. A method comprising:

receiving, by a computing device and in association with a dialogue process with a user, natural language input;

parsing the natural language input to determine;

a first portion of the natural language input corresponding to a first natural language understanding (NLU) model, a second portion of the natural language input corresponding to a second NLU model, and a third portion of the natural language input different from the first and second portions;

processing, using the first NLU model, the first portion of the natural language input;

processing, using the second NLU model, the second portion of the natural language input;

processing, using a third NLU model, the third portion of the natural language input;

determining a semantic interpretation corresponding to the natural language input; and

generating, by the computing device and in association with the dialogue process with the user, output, wherein the output is based on the semantic interpretation.

2. The method of claim 1, wherein:

the first NLU model comprises a first specialized model;

the second NLU model comprises a second specialized model; and

the third NLU model comprises a generic model.

3. The method of claim 1, further comprising determining that a specialized NLU model is available for the first portion of the natural language input.

4. The method of claim 1, further comprising determining that a specialized NLU model is available for the second portion of the natural language input.

5. The method of claim 1, further comprising identifying, based on the natural language input, the first NLU model.

6. The method of claim 1, further comprising identifying, based on the natural language input, the second NLU model.

7. The method of claim 1, wherein the third portion corresponds to neither the first NLU model nor the second NLU model.

8. The method of claim 1, further comprising:

determining a first interpretation of the first portion of the natural language input;

determining a second interpretation of the first portion of the natural language input; and

comparing, using the first NLU model, the first interpretation and the second interpretation.

9. The method of claim 8, further comprising selecting, based on the comparing, the first interpretation or the second interpretation.

10. A method comprising:

receiving, by a computing device and in association with a dialogue process with a user, natural language input;

parsing the natural language input to determine;

a first portion of the natural language input corresponding to a first type of input, and a second portion of the natural language input that does not correspond to the first type of input;

processing, using a first natural language understanding (NLU) model corresponding to the first type of input, the first portion of the natural language input;

processing, using a second NLU model, the second portion of the natural language input; and

generating, by the computing device and in association with the dialogue process with the user, output, wherein the output is based on the natural language input.

11. The method of claim 10, wherein:

the first NLU model comprises a specialized model; and

the second NLU model comprises a generic model.

12. The method of claim 10, further comprising:

determining a first interpretation of the first portion of the natural language input;

determining a second interpretation of the first portion of the natural language input; and

selecting, using the first NLU model, the first interpretation or the second interpretation.

13. The method of claim 10 further comprising identifying, based on the natural language input, the first NLU model.

14. The method of claim 10, wherein the natural language input comprises speech input.

15. The method of claim 10, further comprising determining that the second portion does not correspond to any one of a plurality of specialized models.

16. An apparatus comprising:

at least one processor, and memory storing instructions that, when executed by the at least one processor, cause the apparatus to;

receive, in association with a dialogue process with a user, natural language input;

parse the natural language input to determine;

a first portion of the natural language input corresponding to a first natural language understanding (NLU) model, and a second portion of the natural language input corresponding to a second NLU model;

process, using the first NLU model, the first portion of the natural language input;

process, using the second NLU model, the second portion of the natural language input;

determine a semantic interpretation corresponding to the natural language input; and

generate, in association with the dialogue process with the user, output, wherein the output is based on the semantic interpretation.

17. The apparatus of claim 16, wherein:

the first NLU model comprises a first specialized model; and

the second NLU model comprises a second specialized model.

18. The apparatus of claim 16, wherein the instructions, when executed by the at least one processor, cause the apparatus to:

determine a first interpretation of the first portion of the natural language input;

determine a second interpretation of the first portion of the natural language input; and

select, using the first NLU model, the first interpretation or the second interpretation.

19. The apparatus of claim 16, wherein the instructions, when executed by the at least one processor, cause the apparatus to:

parse the natural language input to determine a third portion of the natural language input different from the first and second portions;

determine a first interpretation of the third portion of the natural language input;

determine a second interpretation of the third portion of the natural language input; and

select, using a generic NLU model, the first interpretation or the second interpretation.

20. The apparatus of claim 16, wherein the natural language input comprises received audio input.