Identifying and analyzing actions from vector representations of alphanumeric characters using a large language model

Patent metadata

Field	Value
Publication number	US-12197859-B1
Application number	US-202418782019-A
Country	US
Kind code	B1
Filing date	Jul 23, 2024
Priority date	Apr 11, 2024
Publication date	Jan 14, 2025
Grant date	Jan 14, 2025

Abstract

The systems and methods disclosed herein receive an output generation request from that includes input for generating an output using a language model. The input includes a set of alphanumeric characters associated with operative standards for a first set of actions. The system divides the set of alphanumeric characters into text subsets. For each text subset, a vector representation is determined. Prompts are created for each vector representation including the set of alphanumeric characters, query contexts, keywords, and/or the text subset. Each vector representation's prompt is input into the language model, which generates a second set of actions of related actions, where subsequently generated actions are based on prior generated actions. The system aggregates the second set of actions into a third set of actions and displays a graphical layout. The graphical layout displays a representation of the set of alphanumeric characters and the corresponding actions.

First claim

We claim:

1. A non-transitory computer-readable storage medium comprising instructions stored thereon, wherein the instructions when executed by at least one data processor of a system, cause the system to:
receive, from a user interface, an output generation request including an input for generation of an output using a language model,
wherein the input includes a set of alphanumeric characters representing operative standards associated with a first set of actions configured to adhere to constraints of the set of alphanumeric characters;

partition the set of alphanumeric characters into a plurality of text subsets of the set of alphanumeric characters using predetermined criteria,
wherein the predetermined criteria includes one or more of: a length of each text subset or a complexity of each text subset;

for each text subset in the plurality of text subsets, identify a set of vector representations representative of the text subset;
using the set of vector representations, construct a set of prompts for each vector representation in the set of vector representations,
wherein each prompt of the set of prompts for a particular vector representation includes two or more of: (1) the set of alphanumeric characters, (2) a set of pre-loaded query context defining the first set of actions, (3) a set of keywords associated with the set of alphanumeric characters of corresponding vector representations of the first set of actions, or (4) the text subset;

for each vector representation in the set of vector representations, map each vector representation to one or more actions in the first set of actions by:
supplying one or more prompts of the set of prompts of the particular vector representation into the language model, and
responsive to supplying the one or more prompts of the set of prompts, receiving, from the language model, a generated second set of actions comprising the one or more actions associated with the particular vector representation,
wherein subsequent actions in the second set of actions generated subsequent to previous actions in the second set of actions are generated using the previous actions in the second set of actions;

generate a third set of actions by aggregating the second sets of actions of the set of text subsets; and
generate for display at the user interface, a layout indicating the third set of actions associated with the set of text subsets, wherein the layout includes a first representation of the set of alphanumeric characters and a second representation of corresponding actions in the third set of actions.

2. The non-transitory, computer-readable storage medium of claim 1 , wherein the instructions further cause the system to:
provide a fourth set of actions associated with current operative standards; and
identify a set of gaps by comparing the third set of actions with the fourth set of actions,
wherein the set of gaps include actions present in the third set of actions and absent in the fourth set of actions.

3. The non-transitory, computer-readable storage medium of claim 2 , wherein the instructions further cause the system to:
determine, for each gap in the set of gaps, a type of difference between a corresponding action within the third set of actions and the fourth set of actions,
wherein the type of difference is associated with one or more of: a presence of a particular set of alphanumeric characters within the actions associated with current operative standards or a semantic meaning of one or more sets of alphanumeric characters within the actions associated with current operative standards; and

using the type of difference, classify each gap in the set of gaps.

4. The non-transitory, computer-readable storage medium of claim 3 , wherein classifying each gap in the set of gaps includes generating one or more of:
a binary indicator of a severity level of the gap,
a category of the severity level of the gap, or
a probability associated with the severity level of the gap.

5. The non-transitory, computer-readable storage medium of claim 3 , wherein the instructions further cause the system to:
in response to the classification of a particular gap in the set of gaps reaching a predetermined threshold, add the corresponding action within the third set of actions to the fourth set of actions.

6. The non-transitory, computer-readable storage medium of claim 5 , wherein the instructions further cause the system to:
using the added corresponding action, automatically triggering an automated workflow indicated by the added corresponding action,
wherein the automated workflow includes adding a representation of the corresponding action to the current operative standards.

7. The non-transitory, computer-readable storage medium of claim 1 , wherein each vector representation in the set of vector representations include one or more electronic tags associated with a type of operation, wherein the instructions further cause the system to:
receive, from the user interface, the type of operation; and
using the type of operation, obtaining a relevant set of alphanumeric characters, via an Application Programming Interface (API).

8. A method performed by a computer device for mapping guidelines to one or more actions, the method comprising:
receiving an output generation request including an input for generation of an output using an AI model,
wherein the input includes a set of alphanumeric characters representing operative standards associated with a first set of actions configured to adhere to constraints of the set of alphanumeric characters;

partition the set of alphanumeric characters into a plurality of text subsets of the set of alphanumeric characters using predetermined criteria,
wherein the predetermined criteria includes one or more of: a length of each text subset or a complexity of each text subset;

for each text subset in the plurality of text subsets, identify a set of vector representations representative of the text subset;
using the set of vector representations, construct a set of prompts for each vector representation in the set of vector representations,
wherein each prompt of the set of prompts for a particular vector representation includes two or more of: (1) the set of alphanumeric characters, (2) a set of pre-loaded query context defining the first set of actions, (3) a set of keywords associated with the set of alphanumeric characters of corresponding vector representations of the first set of actions, or (4) the text subset;

for each vector representation in the set of vector representations, mapping each vector representation to one or more actions in the first set of actions by:
supplying one or more prompts of the set of prompts of the particular vector representation into the AI model, and
responsive to supplying the one or more prompts of the set of prompts, receiving, from the AI model, a generated second set of actions comprising the one or more actions associated with the particular vector representation,
wherein subsequent actions in the second set of actions generated subsequent to previous actions in the second set of actions are generated using the previous actions in the second set of actions; and

generate a third set of actions by aggregating the second sets of actions of the set of text subsets.

9. The method of claim 8 , wherein the AI model is a first AI model, further comprising:
supplying at least a portion of the set of alphanumeric characters into a second AI model; and
receiving from the second AI model, a set of summaries summarizing the set of alphanumeric characters,
wherein at least one prompt in the set of prompts includes one or more summaries in the set of summaries.

10.

Assignees

• Citibank Na

Inventors

• Malviya Shardul
• Liao Wayne
• Jain Deepak
• Cory Samantha
• Saternus Mariusz
• Lewandowski Daniel
• Rath Biraj Krushna
• Murray Stuart
• Davies Philip
• Jain Payal
• Maonah Tariq Husayn

Classifications

• G06F40/216 Primary

  using statistical methods · CPC title