Granted Patent B2

US 12,639,348 · App. 19/286,971 · Granted May 26, 2026

Inventors: Vijay Madisetti (Alpharetta, GA); Arshdeep Bahga (Chandigarh, IN)

Assignee: Vijay Madisetti

G06F16/3329G06F40/284

Abstract

Claims (127)

1 . A method for generating targeted advertisements in large language model (LLM) systems comprising:

receiving a user query at an AI input broker from a user;

identifying one or more categories of information by analyzing the user query using one or more topic modeling techniques;

generating one or more derived queries from the user query to identify at least one of a user intention and a user attitude;

generating a plurality of query responses by processing the user query and the one or more derived queries through one or more families of LLMs (h-LLMs);

determining advertisement content based on the one or more categories of information and based on at least one of the user intention or the user attitude;

generating one or more targeted advertisements responsive to the advertisement content; and

creating an advertisement-enhanced response by integrating the one or more targeted advertisements with the plurality of query responses.

2 . The method of claim 1 wherein determining advertisement content comprises:

implementing a bidding system for advertisers to target the one or more categories of information;

selecting winning advertisement content from the bidding system based on at least one of bid amounts and relevance scores; and

associating the winning advertisement content with a corresponding category of information of the one or more categories of information.

3 . The method of claim 1 wherein generating one or more derived queries comprises:

generating one or more clarifying questions to be presented to the user;

transmitting the one or more clarifying questions to a user device associated with the user;

receiving one or more user responses to the one or more clarifying questions;

analyzing the one or more user responses to determine specific user goals and intentions; and

creating one or more targeted derived queries based on the user goals and intentions.

4 . The method of claim 3 wherein the one or more clarifying questions comprises queries about user purchasing intentions within a predetermined time period for products or services related to the one or more categories.

5 . The method of claim 1 wherein:

a plurality of documents stored on at least one document database comprises a plurality of superchunks; and

integrating the one or more targeted advertisements comprises:

inserting one or more content-based advertisements comprised by the one or more targeted advertisements into a superchunk of the plurality of superchunks based on a topical content of the superchunk of the plurality of superchunks;

linking one or more category-specific advertisements comprised by the one or more targeted advertisements to a classification category of the superchunk of the plurality of superchunks; and

associating each of the one or more category-specific advertisements with the plurality of query responses based on relevance matching scores.

6 . The method of claim 1 further comprising:

maintaining user interaction history data;

identifying at least one of user preferences or behavioral patterns by analyzing the user interaction history data;

adjusting the one or more targeted advertisements based on the at least one of user preferences or behavioral patterns; and

personalizing the advertisement-enhanced response based on user subscription tier information associated with the user.

7 . The method of claim 1 wherein the AI input broker is configured to:

generate follow-up questions responsive to the user query;

transmit the follow-up questions to the user;

receive user responses to the follow-up questions;

generate at least one of predicted future user actions or predicted future purchasing behavior by analyzing the user responses; and

generate predictive advertisement content based on the at least one of predicted future user actions or predicted future purchasing behavior.

8 . A system for generating targeted advertisements in large language model (LLM) systems comprising:

a processor;

a communication device operably coupled to the processor and operable to transmit and receive messages across a digital network; and

a non-transitory computer-readable storage medium having software stored thereon that, when executed by the processor, is operable to:

operate an AI input broker operable to receive a user query;

identify one or more categories of information by analyzing the user query using one or more topic modeling techniques;

generate one or more derived queries from the user query to identify at least one of a user intention and a user attitude;

generate a plurality of query responses by processing the user query and the one or more derived queries through one or more families of LLMs (h-LLMs);

determine advertisement content based on the one or more categories of information and based on at least one of the user intention or the user attitude;

generate one or more targeted advertisements responsive to the advertisement content; and

create an advertisement-enhanced response by integrating the one or more targeted advertisements with the plurality of query responses.

9 . The system of claim 8 wherein the software, when executed by the processor, is operable to determine the advertisement content by:

implementing a bidding system for advertisers to target the one or more categories of information;

selecting winning advertisement content from the bidding system based on at least one of bid amounts and relevance scores; and

associating the winning advertisement content with a corresponding category of information of the one or more categories of information.

10 . The system of claim 8 wherein the software, when executed by the processor, is operable to generate one or more derived queries by:

generating one or more clarifying questions to be presented to the user;

transmitting the one or more clarifying questions to a user device associated with the user;

receiving one or more user responses to the one or more clarifying questions;

analyzing the one or more user responses to determine specific user goals and intentions; and

creating one or more targeted derived queries based on the user goals and intentions.

11 . The system of claim 10 wherein the one or more clarifying questions comprises queries about user purchasing intentions within a predetermined time period for products or services related to the one or more categories.

12 . The system of claim 8 further comprising a document database comprising a plurality of documents that comprises a plurality of superchunks; wherein the software, when executed by the processor, is operable to integrate the one or more targeted advertisements by:

inserting one or more content-based advertisements comprised by the one or more targeted advertisements into a superchunk of the plurality of superchunks based on a topical content of the superchunk of the plurality of superchunks;

linking one or more category-specific advertisements comprised by the one or more targeted advertisements to a classification category of the superchunk of the plurality of superchunks; and

associating each of the one or more category-specific advertisements with the plurality of query responses based on relevance matching scores.

13 . The system of claim 8 wherein the software, when executed by the processor, is operable to:

maintain user interaction history data;

identify at least one of user preferences or behavioral patterns by analyzing the user interaction history data;

adjust the one or more targeted advertisements based on the at least one of user preferences or behavioral patterns; and

personalize the advertisement-enhanced response based on user subscription tier information associated with the user.

14 . The system of claim 8 wherein the AI input broker is further configured to:

generate follow-up questions responsive to the user query;

transmit the follow-up questions to the user;

receive user responses to the follow-up questions;

generate at least one of predicted future user actions or predicted future purchasing behavior by analyzing the user responses; and

generate predictive advertisement content based on the at least one of predicted future user actions or predicted future purchasing behavior.

15 . A system for generating targeted advertisements in large language model (LLM) systems comprising:

means for operating an AI input broker operable to receive a user query from a user;

means for identifying one or more categories of information by analyzing the user query using one or more topic modeling techniques;

means for generating one or more derived queries from the user query to identify at least one of a user intention and a user attitude;

means for generating a plurality of query responses by processing the user query and the one or more derived queries through one or more families of LLMs (h-LLMs);

means for determining advertisement content based on the one or more categories of information and based on at least one of the user intention or the user attitude;

means for generating one or more targeted advertisements responsive to the advertisement content; and

means for creating an advertisement-enhanced response by integrating the one or more targeted advertisements with the plurality of query responses.

16 . The system of claim 15 wherein the means for determining advertisement content is operable to:

implement a bidding system for advertisers to target the one or more categories of information;

select winning advertisement content from the bidding system based on at least one of bid amounts and relevance scores; and

associate the winning advertisement content with a corresponding category of information of the one or more categories of information.

17 . The system of claim 15 wherein the means for generating one or more derived queries is operable to:

generate one or more clarifying questions to be presented to the user;

transmit the one or more clarifying questions to a user device associated with the user;

receive one or more user responses to the one or more clarifying questions;

analyze the one or more user responses to determine specific user goals and intentions; and

create one or more targeted derived queries based on the user goals and intentions.

18 . The system of claim 17 wherein the one or more clarifying questions comprises queries about user purchasing intentions within a predetermined time period for products or services related to the one or more categories.

19 . The system of claim 15 further comprising:

a document database comprising a plurality of documents that comprises a plurality of superchunks; and

means for integrating the one or more targeted advertisements that is operable to:

insert one or more content-based advertisements comprised by the one or more targeted advertisements into a superchunk of the plurality of superchunks based on a topical content of the superchunk of the plurality of superchunks;

link one or more category-specific advertisements comprised by the one or more targeted advertisements to a classification category of the superchunk of the plurality of superchunks; and

associate each of the one or more category-specific advertisements with the plurality of query responses based on relevance matching scores.

20 . The system of claim 15 further comprising:

means for maintaining user interaction history data;

means for identifying at least one of user preferences or behavioral patterns by analyzing the user interaction history data;

means for adjusting the one or more targeted advertisements based on the at least one of user preferences or behavioral patterns; and

means for personalizing the advertisement-enhanced response based on user subscription tier information associated with the user.

21 . The system of claim 15 wherein the AI input broker is configured to:

generate follow-up questions responsive to the user query;

transmit the follow-up questions to the user;

receive user responses to the follow-up questions;

generate at least one of predicted future user actions or predicted future purchasing behavior by analyzing the user responses; and

generate predictive advertisement content based on the at least one of predicted future user actions or predicted future purchasing behavior.

22 . A system for generating targeted advertisements in LLM systems comprising:

an AI input broker configured to:

receive a user query from a user;

identify at least one intention identification from the user query; and

generate one or more derived queries based on the at least one intention identification;

a topic modeling engine configured to identify one or more advertisement-relevant categories relevant to the user query;

an advertisement generation module configured to create one or more targeted advertisements based on an advertisement-relevant category of the one or more advertisement-relevant categories and an intention identification of the at least one intention identification;

a bidding system interface configured to receive a plurality of advertisement bids from a plurality of advertisers for a plurality of content categories comprising the one or more advertisement-relevant categories; and

an advertisement integration module configured to combine the one or more targeted advertisements with one or more LLM-generated responses generated responsive to the user query.

23 . The system of claim 22 wherein the advertisement generation module is further configured to:

generate one or more content-based advertisements for insertion into superchunks comprised by one or more documents of a plurality of documents;

create one or more category-linked advertisements associated with a content category comprised by the plurality of content categories; and

implement one or more tiered advertisement strategies based on a user authentication level associated with the user and a subscription tier associated with the user.

24 . The system of claim 22 further comprising a user intention prediction engine configured to:

analyze a plurality of user query patterns and an interaction history associated with the user;

generate one or more predictive models for a user purchasing behavior associated with the user;

create a plurality of personalized advertisement recommendations based on the one or more predictive models; and

adjust one or more advertisement targeting parameters associated with the user based on one or more predicted user actions generated by the one or more predictive models.

Assignments (1)

Continuity (18)

References Cited (11)