Cognitive analysis for speech recognition using multi-language vector representations

Patent metadata

Field	Value
Publication number	US-11557284-B2
Application number	US-202016733868-A
Country	US
Kind code	B2
Filing date	Jan 3, 2020
Priority date	Jan 3, 2020
Publication date	Jan 17, 2023
Grant date	Jan 17, 2023

Abstract

A method, system and computer program product for speech recognition using multiple languages includes receiving, by one or more processors, an input from a user, the input includes a sentence in a first language. The one or more processors translate the sentence to a plurality of languages different than the first language, and create vectors associated with the plurality of languages, each vector includes a representation of the sentence in each of the plurality of languages. The one or more processors calculate eigenvectors for each vector associated with a language in the plurality of languages, and based on the calculated eigenvectors, a score is assigned to each of the plurality of languages according to a relevance for determining a meaning of the sentence.

First claim

What is claimed is:

1. A computer-implemented method for speech recognition, comprising:
receiving, by one or more processors, an input from a user, the input including a query associated with a user's intent, the query including a sentence in a first language;
determining, by the one or more processors, a degree of similarity between the sentence in the first language and at least one another sentence in the first language;
responsive to the sentence in the first language being written similarly to the at least one another sentence in the first language, encoding, by the one or more processors, the sentence in the first language to generate a plurality of intermediate vectors representing a meaning of the sentence in the first language;
decoding, by the one or more processors, each intermediate vector to a plurality of languages different from the first language, each decoded intermediate vector including a semantic representation of the sentence in each of the plurality of languages for determining the meaning of the sentence in the first language that matches the user's intent in the query;
calculating, by the one or more processors, eigenvectors for each decoded intermediate vector associated with a language in the plurality of languages;
based on the calculated eigenvectors, assigning, by the one or more processors, a score to each of the plurality of languages according to a relevance for determining the meaning of the sentence in the first language that matches the user's intent;
selecting, by the one or more processors, a sentence in a language from the plurality of languages having a highest score for determining the meaning of the sentence in the first language that matches the user's intent; and
based on a meaning of the selected sentence matching the user's intent, using, by the one or more processors, the sentence with the highest score to identify the meaning of the sentence in the first language matching the user's intent, and communicating the meaning of the sentence in the first language to a user device.

2. The method of claim 1 , wherein the selected sentence in the language from the plurality of languages with the highest score is semantically similar to the sentence in the first language.

3. The method of claim 2 , wherein the selected sentence in the language from the plurality of languages with the highest score matches an intent of the sentence.

4. The method of claim 1 , wherein the sentence in the first language is written similarly to a second sentence in the same language but has a different meaning.

5. The method of claim 1 , wherein the sentence in the first language translated to a second language selected from the plurality of languages is written in a different way than the sentence in the first language but are semantically similar.

6. The method of claim 1 , wherein the creating vectors associated with the plurality of languages is done using an encoder-decoder method.

7. A computer system for intent recognition, comprising:
one or more processors, one or more computer-readable memories, one or more computer-readable tangible storage devices, and program instructions stored on at least one of the one or more storage devices for execution by at least one of the one or more processors via at least one of the one or more memories, wherein the computer system is capable of performing a method comprising:
receiving, by one or more processors, an input from a user, the input including a query associated with a user's intent, the query including a sentence in a first language;
determining, by the one or more processors, a degree of similarity between the sentence in the first language and at least one another sentence in the first language;
responsive to the sentence in the first language being written similarly to the at least one another sentence in the first language, encoding, by the one or more processors, the sentence in the first language to generate a plurality of intermediate vectors representing a meaning of the sentence in the first language;
decoding, by the one or more processors, each intermediate vector to a plurality of languages different from the first language, each decoded intermediate vector including a semantic representation of the sentence in each of the plurality of languages for determining the meaning of the sentence in the first language that matches the user's intent in the query;
calculating, by the one or more processors, eigenvectors for each decoded intermediate vector associated with a language in the plurality of languages;
based on the calculated eigenvectors, assigning, by the one or more processors, a score to each of the plurality of languages according to a relevance for determining the meaning of the sentence in the first language that matches the user's intent;
selecting, by the one or more processors, a sentence in a language from the plurality of languages having a highest score for determining the meaning of the sentence in the first language that matches the user's intent; and
based on a meaning of the selected sentence matching the user's intent, using, by the one or more processors, the sentence with the highest score to identify the meaning of the sentence in the first language matching the user's intent, and communicating the meaning of the sentence in the first language to a user device.

8. The computer system of claim 7 , wherein the selected sentence in the language from the plurality of languages with the highest score is semantically similar to the sentence in the first language.

9. The computer system of claim 8 , wherein the selected sentence in the language from the plurality of languages with the highest score matches an intent of the sentence.

10. The computer system of claim 7 , wherein the sentence in the first language is written similarly to a second sentence in the same language but has a different meaning.

11. The computer system of claim 7 , wherein the sentence in the first language translated to a second language selected from the plurality of languages is written in a different way than the sentence in the first language but are semantically similar.

12. The computer system of claim 7 , wherein the creating vectors associated with the plurality of languages is done using an encoder-decoder method.

13. A computer program product for intent recognition, comprising:
a computer readable storage medium having program instructions embodied therewith, wherein the computer readable storage medium is not a transitory signal per se, the program instructions executable by a device to cause the device to perform a method comprising:
receiving, by one or more processors, an input from a user, the input including a query associated with a user's intent, the query including a sentence in a first language;
determining, by the one or more processors, a degree of similarity between the sentence in the first language and at least one another sentence in the first language;
responsive to the sentence in the first language being written similarly to the at least one another sentence in the first language, encoding, by the one or more processors, the sentence in the first language to generate a plurality of intermediate vectors representing a meaning of the sentence in the first language;
decoding, by the one or more processors, each intermediate vector to a plurality of languages different from the first language, each decoded intermediate vector including a semantic representation of the sentence in each of the plurality of languages for determining the meaning of the sentence in the first language that matches the user's intent in the query;
calculating, by the one

Assignees

• IBM

Inventors

• Yuan Zhong Fang
• Yin kun yan
• Li He
• Liu Tong
• Ji Hai

Classifications

• G10L19/00

  Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis (in musical instruments G10H) · CPC title

• G10L15/005

  Language recognition · CPC title

• G06F40/30

  Semantic analysis · CPC title

• G10L15/1815 Primary

  Semantic context, e.g. disambiguation of the recognition hypotheses based on word meaning · CPC title

• G10L15/1822 Primary

  Parsing for meaning understanding · CPC title