Granted Patent B2

US 12,400,084 · App. 17/578,980 · Granted Aug 26, 2025

Inventors: Ahmed Nizam Mohaideen P (Kovilpatti, IN); Sashikumar Venkataraman (Andover, MA); Manik Malhotra (Bengaluru KA, IN)

Assignee: Adeia Guides Inc.

G06F40/35G06F16/243G06F16/248G06F16/3325G06F16/3334G06F16/3338G06F16/9535G06F40/284G06N3/02G10L15/16

Abstract

Claims (96)

1. A method comprising:

receiving a first query and a second query, wherein the first query is received prior to receiving the second query, and wherein the first and second query are received via a user input interface of a device, and the second query is a current query for which a response has not yet been provided;

inputting the first query, and the second query as inputs of a neural network trained on a training dataset comprising:

(i) multiple previous-next query pairs where context is preserved; and

(ii) multiple previous-next query pairs where context is not preserved;

to produce an output result indicating either existence of context change from the first query to the second query or no existence of context change from the first query to the second query, wherein producing the output is further based at least in part on:

generating a first set of tokens based on terms in the first query and a second set of tokens based on terms in the second query;

mapping the first set of tokens and the second set of tokens to a first set of nodes;

comparing a token from the first set of tokens to a set of filler words to determine whether the token matches a filler word of the set of filler words; and

based at least in part on determining that the token matches the filler word of the set of filler words, excluding the token from the first set of tokens;

receiving the output result from the neural network;

determining whether the output result from the neural network indicates existence of context change from the first query to the second query, wherein:

determining that the output result from the neural network indicates no existence of context change from the first query to the second query causes:

automatic generation of a new query by merging the first query and the second query; and

determining that the neural network indicates existence of context change from the first query to the second query causes:

automatic selection of a first portion of the first query and a second portion of the second query that correspond to each other; and

automatic generation of a new query by replacing the first portion of the first query with the second portion of the second query.

2. The method of claim 1 , wherein the neural network comprises the first set of nodes associated with an input layer of the neural network and a second set of nodes associated with a hidden layer of the neural network, further comprising:

training the neural network, based on a training data set, to determine weights associated with connections between the first set of nodes and the second set of nodes in the neural network.

3. The method of claim 2 , wherein determining whether the first query and the second query are associated with the result indicating a merge or a replace operation comprises:

determining, using the weights associated with the connections between the first set of nodes and the second set of nodes, a value indicating whether the first query and the second query are associated with the merge operation or the replace operation.

4. The method of claim 1 , wherein each node of the first set of nodes is associated with a token, and wherein mapping the first set of tokens to the first set of nodes comprises:

matching a first token of the first set of tokens to a token associated with a first node of the first set of nodes of the input layer; and

in response to the matching, updating a first value in the neural network associated with the first node to indicate that a token associated with the first node matches the first token.

5. The method of claim 4 , wherein determining whether the first query and the second query are associated with the result indicating a merge or a replace operation comprises:

retrieving weights associated with the connections between the first set of nodes and a second set of nodes;

determining a first set of values each associated with a respective node of the second set of nodes based on multiplying a second set of values each associated with a respective node of the first set of nodes by the weights associated with the connections between the first set of nodes and the second set of nodes; and

determining a value indicating whether the first query and the second query are associated with the result indication the merge or the replace operation by multiplying the second set of values by the weights associated with the connections between the second set of nodes and the node associated with the value and adding the resulting values.

6. The method of claim 1 , wherein the first query and the second query are received via a voice recognition interface, further comprising, converting the first query to a first string of words based on a speech-to-text conversion and converting the second query to a second string of words based on the speech-to-text conversion.

7. The method of claim 1 , further comprising:

based on determining that the output result from the neural network indicates no existence of context change from the first query to the second query, automatically generating a new query by merging the first query and the second query.

8. The method of claim 1 , further comprising:

based on determining that the neural network indicates existence of context change from the first query to the second query:

automatically selecting a first portion of the first query and a second portion of the second query that correspond to each other; and

automatically generating a new query by replacing the first portion of the first query with the second portion of the second query.

9. The method of claim 1 , wherein when the automatic merging is performed, the method further comprises:

identifying a first entity of the first query;

identifying a second entity of the second query;

in a knowledge graph comprising a first node representing the first entity and a second node representing the second entity, determining that each of the first node and the second node is connected to node representing a specific entity type, and

wherein the inputting further comprises inputting an indication of the specific entity type to the neural network.

10. The method of claim 9 , wherein the first entity corresponds to the first token and the second entity corresponds to the second token.

11. A system comprising:

input/output circuitry configured to:

receive a first query and a second query, wherein the first query is received prior to receiving the second query, and wherein the first and second query are received via a user input interface of a device, and the second query is a current query for which a response has not yet been provided;

processing circuitry configured to:

input the first query and the second query as inputs of a neural network trained on a training dataset comprising:

(i) multiple previous-next query pairs where context is preserved; and

(ii) multiple previous-next query pairs where context is not preserved;

to produce an output result indicating either existence of context change from the first query to the second query or no existence of context change from the first query to the second query, wherein the processing circuitry is configured to produce the output at least in part by:

generating a first set of tokens based on terms in the first query and a second set of tokens based on terms in the second query;

mapping the first set of tokens and the second set of tokens to a first set of nodes;

comparing a token from the first set of tokens to a set of filler words to determine whether the token matches a filler word of the set of filler words; and

based at least in part on determining that the token matches the filler word of the set of filler words, excluding the token from the first set of tokens;

receive the output result from the neural network;

determine whether the output result from the neural network indicates existence of context change from the first query to the second query, wherein:

determining that the output result from the neural network indicates no existence of context change from the first query to the second query causes the processing circuitry to:

automatically generate a new query by merging the first query and the second query; and

determining that the neural network indicates existence of context change from the first query to the second query causes the processing circuitry to:

automatically select a first portion of the first query and a second portion of the second query that correspond to each other; and

automatically generate a new query by replacing the first portion of the first query with the second portion of the second query.

12. The system of claim 11 , wherein the neural network comprises the first set of nodes associated with an input layer of the neural network and a second set of nodes associated with a hidden layer of the neural network, and wherein the processing circuitry is further configured to:

train the neural network, based on a training data set, to determine weights associated with connections between the first set of nodes and the second set of nodes in the neural network.

13. The system of claim 12 , wherein the processing circuitry is further configured, when determining whether the first query and the second query are associated with the result indicating a merge operation and a replace operation, to:

determine, using the weights associated with the connections between the first set of nodes and the second set of nodes, a value indicating whether the first query and the second query are associated with the merge operation and the replace operation.

14. The system of claim 11 , wherein each node of the first set of nodes is associated with a token, and wherein the processing circuitry is further configured, when mapping the first set of tokens to the first set of nodes, to:

match a first token of the first set of tokens to a token associated with a first node of the first set of nodes of the input layer; and

in response to the matching, update a first value in the neural network associated with the first node to indicate that a token associated with the first node matches the first token.

15. The system of claim 14 , wherein the processing circuitry is further configured, when determining whether the first query and the second query are associated with the result indicating a merge operation and a replace operation, to:

retrieve weights associated with the connections between the first set of nodes and a second set of nodes;

determine a first set of values each associated with a respective node of the second set of nodes based on multiplying a second set of values each associated with a respective node of the first set of nodes by the weights associated with the connections between the first set of nodes and the second set of nodes; and

determine a value indicating whether the first query and the second query are associated with the result indication the merge operation and the replace operation by multiplying the second set of values by the weights associated with the connections between the second set of nodes and the node associated with the value and adding the resulting values.

16. The system of claim 11 , wherein the processing circuitry is configured to:

based on determining that the output result from the neural network indicates no existence of context change from the first query to the second query, automatically generate a new query by merging the first query and the second query.

17. The system of claim 11 , wherein the processing circuitry is configured to:

based on determining that the neural network indicates existence of context change from the first query to the second query:

automatically select a first portion of the first query and a second portion of the second query that correspond to each other; and

automatically generate a new query by replacing the first portion of the first query with the second portion of the second query.

18. The system of claim 11 , wherein when the automatic merging is performed, the processing circuitry is further configured to:

identify a first entity of the first query;

identify a second entity of the second query;

in a knowledge graph comprising a first node representing the first entity and a second node representing the second entity, determine that each of the first node and the second node is connected to node representing a specific entity type, and

perform the inputting further by inputting an indication of the specific entity type to the neural network.

19. The system of claim 18 , wherein the first entity corresponds to the first token and the second entity corresponds to the second token.

20. A method comprising:

receiving a first query and a second query, wherein the first query is received prior to receiving the second query, and wherein the first and second query are received via a user input interface of a device, and the second query is a current query for which a response has not yet been provided;

prior to receiving the second query, providing a first response to the first query based at least in part on identifying a first connection in a knowledge graph between the first entity of the first query and the first entity type for the first entity;

determining that a second term of the second query corresponds to a second entity type indicated in the knowledge graph; and

identifying a second connection in the knowledge graph between the second entity type and the first entity of the first query;

inputting the first query, the second query, and an indication of the second connection to the neural network as inputs of a neural network trained on a training dataset comprising:

(i) multiple previous-next query pairs where context is preserved; and

(ii) multiple previous-next query pairs where context is not preserved;

to produce an output result indicating either existence of context change from the first query to the second query or no existence of context change from the first query to the second query;

receiving the output result from the neural network;

determining the output result from the neural network indicates existence of context change from the first query to the second query;

based at least in part on determining the output result from the neural network indicates existence of context change from the first query to the second query, automatically updating the first entity type of the first entity which the second query is clarifying; and

providing a second response to the second query based at least in part on the automatic updating of the first entity type, wherein the second response comprises one or more different results than the first response.

Assignments (3)

Continuity (3)

References Cited (31)

Cited By (1)

US 12,694,046