Granted Patent B2

US 12,619,430 · App. 18/379,408 · Granted May 5, 2026

Inventors: Rares Dolga (London, GB); Yulong Pei (Eindhoven, NL); Vali Tawosi (London, GB); Salwa Husam Alamir (Bournemouth, GB); Sameena Shah (Scarsdale, NY)

Assignee: JPMORGAN CHASE BANK, N.A.

G06F8/77G06Q10/063112

Abstract

Claims (58)

1 . A method for evidencing skills of a developer, the method comprising:

extracting, by a processor and from a plurality of databases, a list of terms and corresponding definitions;

removing, by the processor, at least one term from the extracted list of terms and corresponding definitions;

identifying, by the processor, a plurality of skills based on the extracted list of terms and corresponding definitions after the removing of the at least one term;

grouping, by the processor, the identified plurality of skills for forming at least one ontology based on similarity, wherein the grouping is performed using affinity clustering, and wherein the affinity clustering includes extracting a sentence embedding using a language model, the sentence embedding includes a process of representing variable-length sentences as fixed-length vectors;

acquiring, by the processor and from one or more data sources, raw data;

performing, by the processor, data cleaning on the acquired raw data;

identifying, by the processor, at least one domain-specific skill from the cleaned raw data;

performing, by the processor, skill matching by comparing the at least one domain-specific skill against the at least one ontology, wherein a similarity between the at least one domain-specific skill and the at least one ontology is measured based on a cosine of an angle formed between a first embedding vector of the at least one domain-specific skill and a second embedding vector of the at least one ontology to determine whether or not the first embedding vector and the second embedding vector are pointing in a same direction within a deviation below a reference threshold;

when the at least one domain-specific skill matches with the at least one ontology, determining that the at least one ontology is evidenced for the developer;

performing, by the processor, analytics to identify and display, on a display, the evidenced at least one ontology and corresponding level; and

automatically assigning, by the processor, at least one task based on the performed analytics.

2 . The method according to claim 1 , wherein the plurality of databases includes a glossary storage and a document storage.

3 . The method according to claim 1 , wherein the at least one ontology is formed independently from the identifying of the at least one domain-specific skill from the cleaned raw data.

4 . The method according to claim 1 , wherein the one or more data sources include at least a task storage.

5 . The method according to claim 1 , wherein the one or more data sources include at least a message storage.

6 . The method according to claim 1 , wherein the raw data is text data.

7 . The method according to claim 1 , wherein the corresponding level is determined by a number of tasks performed associated with the at least one ontology.

8 . The method according to claim 1 , wherein the at least one ontology includes a plurality of ontologies, and

wherein the corresponding level is displayed for each of the plurality of ontologies.

9 . The method according to claim 8 , wherein the automatic assigning is performed based on highest level for an ontology among the plurality of ontologies.

10 . The method according to claim 8 , wherein the plurality of ontologies and the corresponding levels for the developer are displayed as a singular bar.

11 . The method according to claim 10 , wherein the plurality of ontologies and corresponding levels for the developer are displayed in different colors.

12 . The method according to claim 1 , wherein the data cleaning includes removal of extraneous information or noise.

13 . The method according to claim 1 , further comprising:

automatically recommending, by the processor, a new position based on the evidenced at least one ontology and corresponding level.

14 . The method according to claim 1 , wherein the skill matching includes executing at least two machine learning models, and

wherein the at least two machine learning models includes a rule based model and a large language model.

15 . The method according to claim 14 , wherein the rule based model utilizes the at least one ontology to detect a presence of a skill term corresponding to the at least one ontology, and

wherein the skill term includes an exact term, a synonym, an acronym or a misspelling of the skill term.

16 . The method according to claim 14 , wherein the large language model further processes an output provided by the rule based model, and wherein the large language model utilizes in-context learning to extract a list of skill terms from the raw data.

17 . A system for evidencing skills of a developer, the system comprising:

a memory; and

a processor,

wherein the system is configured to perform:

extracting, from a plurality of databases, a list of terms and corresponding definitions;

removing at least one term from the extracted list of terms and corresponding definitions;

identifying a plurality of skills based on the extracted list of terms and corresponding definitions after the removing of the at least one term;

grouping the identified plurality of skills for forming at least one ontology based on similarity, wherein the grouping is performed using affinity clustering, and wherein the affinity clustering includes extracting a sentence embedding using a language model, the sentence embedding includes a process of representing variable-length sentences as fixed-length vectors;

acquiring, from one or more data sources, raw data;

performing data cleaning on the acquired raw data;

identifying at least one domain-specific skill from the cleaned raw data;

performing skill matching by comparing the at least one domain-specific skill against the at least one ontology, wherein a similarity between the at least one domain-specific skill and the at least one ontology is measured based on a cosine of an angle formed between a first embedding vector of the at least one domain-specific skill and a second embedding vector of the at least one ontology to determine whether or not the first embedding vector and the second embedding vector are pointing in a same direction within a deviation below a reference threshold;

when the at least one domain-specific skill matches with the at least one ontology, determining that the at least one ontology is evidenced for the developer;

performing analytics to identify and display, on a display, the evidenced at least one ontology and corresponding level; and

automatically assigning at least one task based on the performed analytics.

18 . A non-transitory computer readable storage medium that stores a computer program for evidencing skills of a developer, the computer program, when executed by a processor, causing a system to perform a plurality of processes comprising:

extracting, from a plurality of databases, a list of terms and corresponding definitions;

removing at least one term from the extracted list of terms and corresponding definitions;

identifying a plurality of skills based on the extracted list of terms and corresponding definitions after the removing of the at least one term;

grouping the identified plurality of skills for forming at least one ontology based on similarity, wherein the grouping is performed using affinity clustering, and wherein the affinity clustering includes extracting a sentence embedding using a language model, the sentence embedding includes a process of representing variable-length sentences as fixed-length vectors;

acquiring, from one or more data sources, raw data;

performing data cleaning on the acquired raw data;

identifying at least one domain-specific skill from the cleaned raw data;

performing skill matching by comparing the at least one domain-specific skill against the at least one ontology, wherein a similarity between the at least one domain-specific skill and the at least one ontology is measured based on a cosine of an angle formed between a first embedding vector of the at least one domain-specific skill and a second embedding vector of the at least one ontology to determine whether or not the first embedding vector and the second embedding vector are pointing in a same direction within a deviation below a reference threshold;

when the at least one domain-specific skill matches with the at least one ontology, determining that the at least one ontology is evidenced for the developer;

performing analytics to identify and display, on a display, the evidenced at least one ontology and corresponding level; and

automatically assigning at least one task based on the performed analytics.

Assignments (2)

Continuity (1)

References Cited (11)