US 2026/0310659 A1Published application

Systems and methods for automated model generation

AssigneeKobai, Inc.

InventorRyan Oattes

Life of this patent

Estimating the term…

Loading patent…

Abstract

A method for automated model generation is disclosed. The methods and system comprise generating tags from an accessed data source. Each tag may be marked structure class, structure attribute class, record class, or record attribute class. The methods and systems further includes generating (e.g., using an LLM), a vector embedding for each tag, and comparing the generated vector embeddings to generate match confidence levels used to generate different candidate groups. The methods and system also include generating for display a user interface that hierarchically represents each grouping, and adjusting it based on user input. The methods and systems additionally generate a semantic structure comprising classes, attributes, and class links based on the adjusted groupings and provide an interface for answering semantic queries based on the generated semantic structure.

Claims20

Dependent claims

A method comprising:

accessing at least one data source;
extracting metadata from the at least one data source to generate a plurality of tags, wherein each of the plurality of tags is marked with one of a:
structure class, structure attribute class, record class, and record attribute class;
generating, using an LLM, a vector embedding for each tag of the plurality of tags;
generating pairwise comparison data between the generated vector embeddings;
generating a set of candidate record groupings based on the pairwise comparison data, wherein match confidence levels are used to generate different candidate groups for the set of candidate record groupings;
generating for display a user interface that hierarchically represents each grouping of the set of candidate record groupings;
receiving input via the user interface for excluding at least one candidate record grouping from the set of candidate record groupings;
adjusting the candidate record groupings based on the input;
generating a semantic structure comprising classes, attributes, and class links based on the adjusted candidate record groupings;
and
providing an interface for answering semantic queries based on the generated semantic structure.

9 more elements

A system comprising:

control circuitry configured to:
access at least one data source;
extract metadata from the at least one data source to generate a plurality of tags, wherein each of the plurality of tags is marked with one of a:
structure class, structure attribute class, record class, and record attribute class;
generate, using an LLM, a vector embedding for each tag of the plurality of tags;
generate pairwise comparison data between the generated vector embeddings;
generate a set of candidate record groupings based on the pairwise comparison data, wherein match confidence levels are used to generate different candidate groups for the set of candidate record groupings;
I/O circuitry configured to:
generate for display a user interface that hierarchically represents each grouping of the set of candidate record groupings;
receive input via the user interface for excluding at least one candidate record grouping from the set of candidate record groupings;
wherein the control circuitry is further configured to:
adjust the candidate record groupings based on the input;
generate a semantic structure comprising classes, attributes, and class links based on the adjusted candidate record groupings;
and
wherein the I/O circuitry is further configured to:
provide an interface for answering semantic queries based on the generated semantic structure.

13 more elements

Dependencies are inferred from the claim text.