Granted Patent B2

US 12,417,225 · App. 18/809,284 · Granted Sep 16, 2025

Inventors: Jayaprakash Vijayan (Dublin, CA); Ved Surtani (Bengaluru, IN); Nitika Gupta (Bengaluru, IN); Malarvizhi Saravanan (Bengaluru, IN); Anirudh Saria (Kathmandu, NP); Amrutha Dharmaraj (Karnataka, IN)

Assignee: Tekion Corp

G06F16/24522G06F16/215G06F16/2365G06F16/248G06F16/285G06F16/9535G06N20/00

Abstract

Claims (64)

1. A method comprising:

receiving, through a user interface, a first query from a user associated with a vehicle dealership;

applying the first query to a first trained machine learning model configured to predict an intent of the first query;

applying the first query to a second trained machine learning model configured to predict a set of entities of the first query;

generating a normalized representation of the first query based on the predicted intent and the predicted set of entities, wherein a format of the normalized representation of the first query is database language agnostic, and wherein generating the normalized representation comprises filtering out a subset of metadata from an interim representation of the first query that precedes the normalized representation of the first query, the filtering out based on an importance value for each metadata of the subset being below a threshold level of importance;

translating the normalized representation of the first query into a second query having a format compatible with a language of a database of the vehicle dealership;

fetching data from the database of the vehicle dealership associated with the predicted intent and the predicted set of entities using the second query; and

providing, on the user interface, the data for display to the user.

2. The method of claim 1 , wherein the first query is a text query, the method further comprising:

pre-processing the text query using a natural language processing model configured to identify spelling errors in the text query; and

responsive to identifying one or more spelling errors in the text query, modifying the text query to remove the one or more spelling errors.

3. The method of claim 1 , wherein the first query is a voice query, the method further comprising:

receiving audio data from the user via the user interface, the audio data associated with the first query; and

applying the audio data to a third trained machine learning model configured to generate a transcript from the audio data; and

generating a text query from the transcript.

4. The method of claim 1 , wherein the first trained machine learning model is a classification model.

5. The method of claim 4 , wherein the classification model is a supervised machine learning model, and wherein the supervised machine learning model is configured to classify the first query as one or more of an overall intent query, a sales query, or a services query.

6. The method of claim 1 , further comprising: extracting metadata associated with the vehicle dealership, wherein the normalized representation of the first query is further based on the extracted metadata.

7. The method of claim 1 , wherein the first query is associated with a requested action, the method further comprising:

accessing one or more documents associated with the requested action;

accessing metadata associated with the user;

inputting one or more metadata values into the one or more documents based on the requested action; and

providing for display the one or more documents to the user via the user interface.

8. The method of claim 1 , further comprising:

generating a query recommendation for an additional query based on the predicted set of entities of the first query;

providing the query recommendation for the additional query for display on the user interface; and

responsive to receiving an indication that the user interacted with the additional query:

fetching data associated with the additional query; and

providing the data associated with the additional query for display to the user.

9. The method of claim 1 , wherein the second trained machine learning model is an entity determination model, and wherein an entity of a query describes a category of a term of the query.

10. The method of claim 9 , wherein the entity determination model is a deep learning model.

11. The method of claim 1 , wherein generating the normalized representation of the first query further comprises filtering data associated with the first query based on a level of importance associated with one or more data values, and wherein the normalized representation includes data with at least a threshold level of importance.

12. The method of claim 1 , wherein an entity of the predicted set of entities is associated with a first value in a set of values, the method further comprising:

randomizing an additional query recommendation for an additional query based on additional values in the set of values; and

providing the randomized additional query recommendation for the additional query for display on the user interface.

13. The method of claim 1 , further comprising: updating a method of randomizing an additional query recommendation based on one or more responses to additional query recommendations from vehicle dealerships.

14. A non-transitory computer-readable medium comprising memory with instructions encoded thereon, the instructions, when executed, causing one or more processors to perform operations, the instructions comprising instructions to:

receive, through a user interface, a first query from a user associated with a vehicle dealership;

apply the first query to a first trained machine learning model configured to predict an intent of the first query;

apply the first query to a second trained machine learning model configured to predict a set of entities of the first query;

generate a normalized representation of the first query based on the predicted intent and the predicted set of entities, wherein a format of the normalized representation of the first query is database language agnostic, and wherein generating the normalized representation comprises filtering out a subset of metadata from an interim representation of the first query that precedes the normalized representation of the first query, the filtering out based on an importance value for each metadata of the subset being below a threshold level of importance;

translate the normalized representation of the first query into a second query having a format compatible with a language of a database of the vehicle dealership;

fetch data from the database of the vehicle dealership associated with the predicted intent and the predicted set of entities using the second query; and

provide, on the user interface, the data for display to the user.

15. The non-transitory computer-readable medium of claim 14 , wherein the first query is a text query, and wherein the instructions further comprise instructions to:

pre-process the text query using a natural language processing model configured to identify spelling errors in the text query; and

responsive to identifying one or more spelling errors in the text query, modify the text query to remove the one or more spelling errors.

16. The non-transitory computer-readable medium of claim 14 , wherein the first query is a voice query, and wherein the instructions further comprise instructions to:

receive audio data from the user via the user interface, the audio data associated with the first query; and

apply the audio data to a third trained machine learning model configured to generate a transcript from the audio data; and

generate a text query from the transcript.

17. The non-transitory computer-readable medium of claim 14 , wherein the first trained machine learning model is a classification model.

18. The non-transitory computer-readable medium of claim 17 , wherein the classification model is a supervised machine learning model, and wherein the supervised machine learning model is configured to classify the first query as one or more of an overall intent query, a sales query, or a services query.

19. A system comprising:

memory with instructions encoded thereon; and

one or more processors that, when executing the instructions, are caused to perform operations comprising:

receiving, through a user interface, a first query from a user associated with a vehicle dealership;

applying the first query to a first trained machine learning model configured to predict an intent of the first query;

applying the first query to a second trained machine learning model configured to predict a set of entities of the first query;

generating a normalized representation of the first query based on the predicted intent and the predicted set of entities, wherein a format of the normalized representation of the first query is database language agnostic, and wherein generating the normalized representation comprises filtering out a subset of metadata from an interim representation of the first query that precedes the normalized representation of the first query, the filtering out based on an importance value for each metadata of the subset being below a threshold level of importance;

translating the normalized representation of the first query into a second query having a format compatible with a language of a database of the vehicle dealership;

fetching data from the database of the vehicle dealership associated with the predicted intent and the predicted set of entities using the second query; and

providing, on the user interface, the data for display to the user.

20. The system of claim 19 , wherein generating the normalized representation of the first query further comprises filtering data associated with the first query based on a level of importance associated with one or more data values, and wherein the normalized representation includes data with at least a threshold level of importance.

Assignments (2)

Continuity (2)

References Cited (31)