Systems and methods for generation of metadata by an artificial intelligence model based on context

Patent metadata

Field	Value
Publication number	US-12450198-B2
Application number	US-202318497616-A
Country	US
Kind code	B2
Filing date	Oct 30, 2023
Priority date	Oct 30, 2023
Publication date	Oct 21, 2025
Grant date	Oct 21, 2025

Abstract

Disclosed herein are methods and systems for generating metadata from content using one or more machine learning models. In an embodiment, a method may include receiving the content through a graphical user interface associated with the large language model, generating a first file by tokenizing the content into an input format for the large language model and merging the tokenized content with a content instruction, inputting the first file into the large language model, generating, using the large language model, metadata from at least the first file, the metadata reflecting a context associated with the content, generating a second file, the second file comprising the metadata, and displaying the generated metadata on the graphical user interface.

First claim

What is claimed is:

1. A system for generating metadata from content using a large language model, the system comprising:
at least one memory storing instructions; and
at least one processor comprising a plurality of distributed processors configured to execute the instructions to perform operations, wherein the system is configured to perform the operations using load-balancing to maintain stable resource load among the plurality of processors, the operations comprising:
receiving the content through a graphical user interface associated with the large language model;
generating a first file by tokenizing the content into an input format for the large language model and merging the tokenized content with a content instruction;
inputting the first file into the large language model;
generating, using the large language model, a plurality of metadata by determining a main topic from at least the first file, the plurality of metadata reflecting a context associated with the content, wherein the large language model is trained to determine the main topic by identifying a first detail from the first file and removing a second detail from the first file prior to generating the plurality of metadata;
ranking the plurality of metadata according to a relevance of each generated metadata to the first file determined based on a sentiment analysis performed by the large language model, wherein the sentiment analysis performed by the large language model comprises classifying the first file and the generated metadata in one or more emotional attitudes;
determining, based on the ranking, one metadata from the plurality of metadata for display;
generating a second file, the second file comprising the determined one metadata, wherein generating the second file comprises implementing a token constraint and resizing the determined one metadata to comport with a predetermined size, the predetermined size being pre-established for a specific use selected from at least one of thumbnail, title, or abstract; and
displaying the generated second file comprising the determined one metadata on the graphical user interface, wherein displaying the generated second file includes only displaying the determined one metadata of the plurality of metadata.

2. The system of claim 1 , wherein the content comprises at least one of a text query from a user or a text response generated by the large language model.

3. The system of claim 1 , wherein the content comprises at least one of a text document or an image file, and wherein the plurality of metadata comprises a plurality of titles.

4. The system of claim 1 , wherein the at least one processor is further configured to execute the instructions to perform:
generating structured content associated with the received content; and
incorporating the structured content into the content for the input into the large language model.

5. The system of claim 1 , wherein the large language model comprises a multi-modal large language model.

6. The system of claim 5 , wherein the at least one processor is further configured to execute the instructions to perform at least one of:
prompting the large language model, or
fine tuning the large language model.

7. The system of claim 1 , wherein
the content comprises a plurality of words, and
the large language model is trained to generate contextual embeddings for each word of the plurality of words.

8. The system of claim 1 , wherein the large language model is configured to perform abstractive text summarization.

9. The system of claim 1 , wherein the at least one processor is further configured to execute the instructions to perform:
modifying the selected one metadata in the second file by filtering the selected one metadata; and
displaying the generated second file comprises displaying the modified metadata on the graphical user interface.

10. A method for generating metadata from content using a large language model, comprising:
receiving, via at least one processor comprising a plurality of distributed processors configured to use load-balancing to maintain stable resource load among the plurality of processors, the content through a graphical user interface associated with the large language model;
generating, via the at least one processor, a first file by tokenizing the content into an input format for the large language model and merging the tokenized content with a content instruction;
inputting, via the at least one processor, the first file into the large language model;
generating, via the at least one processor using the large language model, a plurality of metadata by determining a main topic from at least the first file, the plurality of metadata reflecting a context associated with the content, wherein the large language model is trained to determine the main topic by identifying a first detail from the first file and removing a second detail from the first file prior to generating the plurality of metadata;
ranking, via the at least one processor, the plurality of metadata according to a relevance of each generated metadata to the first file determined based on a sentiment analysis performed by the large language model, wherein the sentiment analysis performed by the large language model comprises classifying the first file and the generated metadata in one or more emotional attitudes;
determining, via the at least one processor based on the ranking, one metadata from the plurality of metadata for display;
generating, via the at least one processor, a second file, the second file comprising the determined one metadata, wherein generating the second file comprises implementing a token constraint and resizing the determined one metadata to comport with a predetermined size, the predetermined size being pre-established for a specific use selected from at least one of thumbnail, title, or abstract; and
displaying, via the at least one processor, the generated second file comprising the determined one metadata on the graphical user interface, wherein displaying the generated second file includes only displaying the determined one metadata of the plurality of metadata.

11. The method of claim 10 , wherein the content comprises a text query from a user.

12. The method of claim 10 , wherein the content comprises at least one of a text document or an image file, and wherein the plurality of metadata comprises a plurality of titles.

13. The method of claim 10 , further comprising:
generating structured content associated with the received content; and
incorporating the structured content into the content for the input into the large language model.

14. The method of claim 10 , further comprising:
performing at least one of:
prompting the large language model, or
fine tuning the large language model.

15. The method of claim 10 , wherein
the content comprises a plurality of words, and
the large language model is trained to generate contextual embeddings for each word of the plurality of words.

16. The method of claim 10 , further comprising:
modifying the selected one metadata in the second file by filtering the selected one metadata, wherein the modified metadata comprises at least one of a title, subtitle, name, icon, thumbnail, category, summary, description, or image associated with the content; and
displaying the generated second file comprises displaying the modified metadata on the graphical user interface.

17. The system of claim 1 , wherein the relevance of each generated metadata to the first file is further based on a sentiment score calculated by the large lan

Assignees

• Openai Opco Llc
• Openal Opco Llc

Inventors

• Deutsch Noah
• Zweig Benjamin
• Zerfas Valerie
• Simens Madeline
• Heaton Michael
• Turley Nicholas

Classifications

• G06F40/40

  Processing or translation of natural language (natural language analysis G06F40/20; semantic analysis G06F40/30) · CPC title

• G06F40/284

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

• G06F3/048

  Interaction techniques based on graphical user interfaces [GUI] · CPC title

• G06F3/0482

  Interaction with lists of selectable items, e.g. menus · CPC title

• G06F3/04842

  Selection of displayed objects or displayed text elements (G06F3/0482 takes precedence) · CPC title