Method and system for identifying user intent from user statements

Patent metadata

Field	Value
Publication number	US-11144726-B2
Application number	US-201916368899-A
Country	US
Kind code	B2
Filing date	Mar 29, 2019
Priority date	Feb 14, 2019
Publication date	Oct 12, 2021
Grant date	Oct 12, 2021

Abstract

The present disclosure discloses method and a user intent identification system for identifying user intent from user statements. The user intent identification system receives input statement provided by a user from a Natural Language Understanding (NLU) engine. The input statement is processed to remove one or more irrelevant content. A plurality of features for each word in the processed input statement is extracted. The plurality of features comprises Parts of Speech (POS) label, dependency parse tree and word embeddings. The user intent determination system predicts class for each word in the processed input statement from a plurality of predefined classes using a neural network model. The neural network model predicts class for each word based on input vector generated for the each word based on the plurality of features. Thereafter, the user intent is identified based on class predicted for each word in processed input statement.

First claim

What is claimed is:

1. A method for identifying user intent from user statements, the method comprising:
receiving, by a user intent determination system, an input statement provided by a user from a Natural Language Understanding (NLU) engine, wherein the input statement is processed to remove one or more irrelevant content;
extracting, by the user intent determination system, a plurality of features for each word in the processed input statement, wherein the plurality of features comprises a Parts of Speech (POS) label, a dependency parse tree, and word embeddings;
predicting, by the user intent determination system, a class for the each word in the processed input statement from a plurality of predefined classes using a neural network model, wherein the neural network model predicts the class for the each word based on an input vector generated for the each word based on the plurality of features, wherein the input vector comprises the POS label for a target word selected from a plurality of words present in the input statement, the POS label of a predetermined number of words prior to the target word, word embeddings of the target word, a word embedding of a head word in the dependency parse tree and a dependency label for the target word; and
identifying, by the user intent determination system, the user intent based on the class predicted for the each word in the processed input statement, wherein the user intent is provided to the NLU engine to provide a response to the input statement based on the user intent.

2. The method as claimed in claim 1 , wherein the dependency parse tree is generated based on intrinsic dependencies of the each word with each of other words in the processed input statement.

3. The method as claimed in claim 1 , wherein the word embeddings are identified using an artificial word embedding neural network trained using a text corpus of a plurality of natural language sentences, the word embeddings being representation of each of one or more words in a first dimensional vector space.

4. The method as claimed in claim 1 , wherein the neural network model is trained using a text corpus containing a plurality of natural language sentences tagged with the plurality of predefined classes.

5. The method as claimed in claim 1 , wherein the plurality of predefined classes comprise a Begin-Central Idea (BCI), a Inside Central Idea (ICI), a Begin-Central Action (BCA), a Inside-Central Action (ICA), Begin-Central Idea Attributes (BCIA), and an Inside-Central Idea Attributes (ICIA).

6. The method as claimed in claim 1 , wherein the user intent is represented as a central idea, actions and features.

7. A user intent determination system for identifying user intent from user statements, comprising:
a processor; and
a memory communicatively coupled to the processor, wherein the memory stores processor instructions, which, on execution, causes the processor to:
receive an input statement provided by a user from a Natural Language Understanding (NLU) engine, wherein the input statement is processed to remove one or more irrelevant content;
extract a plurality of features for each word in the processed input statement, wherein the plurality of features comprises a Parts of Speech (POS) label, a dependency parse tree and word embeddings;
predict a class for each word in the processed input statement from a plurality of predefined classes using a neural network model, wherein the neural network model predicts the class for each word based on an input vector generated for each word based on the plurality of features, wherein the input vector comprises the POS label for a target word selected from a plurality of words present in the input statement, the POS label of a predetermined number of words prior to the target word, word embeddings of the target word, a word embedding of a head word in the dependency parse tree and a dependency label for the target word; and
identify the user intent based on the class predicted for each word in the processed input statement, wherein the user intent is provided to the NLU engine to provide a response to the input statement based on the user intent.

8. The user intent determination system as claimed in claim 7 , wherein the dependency parse tree is generated based on intrinsic dependencies of the each word with each of other words in the processed input statement.

9. The user intent determination system as claimed in claim 7 , wherein the word embeddings are a representation of the word in a first dimensional vector space, the word embeddings are identified using an artificial word embedding neural network trained using a text corpus of a plurality of natural language sentences.

10. The user intent determination system as claimed in claim 7 , wherein the processor trains the neural network model using a text corpus containing a plurality of natural language sentences tagged with the plurality of predefined classes.

11. The user intent determination system as claimed in claim 7 , wherein the plurality of predefined classes comprises a Begin-Central Idea (BCI), a Inside Central Idea (ICI), a Begin-Central Action (BCA), an Inside-Central Action (ICA), Begin-Central Idea Attributes (BCIA) and an Inside-Central Idea Attributes (ICIA).

12. The user intent determination system as claimed in claim 7 , wherein the user intent is represented as a central idea, actions and features.

13. A non-transitory computer readable medium including instruction stored thereon that when processed by at least one processor cause a user intent determination system to perform operations comprising:
receiving an input statement provided by a user from a Natural Language Understanding (NLU) engine, wherein the input statement is processed to remove one or more irrelevant content;
extracting a plurality of features for each word in the processed input statement, wherein the plurality of features comprises a Parts of Speech (POS) label, a dependency parse tree and word embeddings;
predicting a class for the each word in the processed input statement from a plurality of predefined classes using a neural network model, wherein the neural network model predicts the class for the each word based on an input vector generated for the each word based on the plurality of features, wherein the input vector comprises the POS label for a target word selected from a plurality of words present in the input statement, the POS label of a predetermined number of words prior to the target word, word embeddings of the target word, a word embedding of a head word in the dependency parse tree and a dependency label for the target word; and
identifying the user intent based on the class predicted for the each word in the processed input statement, wherein the user intent is provided to the NLU engine to provide a response to the input statement based on the user intent.

14. The non-transitory computer readable medium as claimed in claim 13 , wherein the dependency parse tree is generated based on intrinsic dependencies of the each word with each of other words in the processed input statement.

15. The non-transitory computer readable medium as claimed in claim 13 , wherein the word embeddings are a representation of the word in a first dimensional vector space, the word embeddings are identified using an artificial word embedding neural network trained using a text corpus of a plurality of natural language sentences.

16. The non-transitory computer readable medium as claimed in claim 13 , wherein the plurality of predefined classes comprises a Begin-Central Idea (BCI), a Inside Central Idea (ICI), a B

Assignees

• Wipro Ltd

Inventors

• Chatterjee Arindam
• Arya Rahul

Classifications

• G06F40/284 Primary

  Lexical analysis, e.g. tokenisation or collocates · CPC title

• G06F40/30 Primary

  Semantic analysis · CPC title

• G06N3/042

  Knowledge-based neural networks; Logical representations of neural networks · CPC title

• G06F18/2431

  Multiple classes · CPC title

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title