# US 2025/0148515 A1 · Artificial intelligence based transactions contextualization platform

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2025/0148515 A1
- Document ID: US20250148515A1
- Status: Published application
- Kind: A1
- Application number: 18/387,799
- Filed: 2023-11-07
- Published: 2025-05-08
- Earliest claimed: 2023-11-07
- Assignees: ExlService Holdings, Inc.
- Applicants: ExlService Holdings, Inc.
- Inventors: Vivek Vinod; Shantanu Paliwal; Abhinav Agarwal; Nikhil Ahuja; SK Praveen
- Web page: https://patentgrep.com/patent/US20250148515A1

## CPC

- G06Q 30/0623: {by investigating goods or services}
- G06N 5/022: Knowledge engineering; Knowledge acquisition
- G06N 7/02: using fuzzy logic (computing arrangements based on biological models G06N3/00; computing arrangements using knowledge-based models G06N5/00)
- G06N 20/00: Machine learning
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A data analytics and contextualization platform instance acquires an input dataset. Using fuzzy logic, the platform generates a set of contextual items based on the unstructured textual data. Items in the input dataset are labeled using contextual items. Using a labeled particular item, a set of entity behavior maps can be generated to include one or more features selected based on a first particular label of the particular item. One or more AI/ML modeling operations can be selected based on a type of the entity behavior map and performed for items in the entity behavior map. A particular first subset of AI/ML modeling operations can be performed, using a particular contextual item as a trigger, in parallel relative to a particular second subset of AI/ML modeling operations. The entity behavior map, which can be populated using outputs of the AI/ML operations, can be transmitted to a target computing system.

## Claims

### Claim 1

A computing system comprising at least one processor, at least one memory, and one or more non-transitory computer readable media storing instructions, which when executed by the at least one processor, perform operations for artificial intelligence/machine learning (AI/ML) based modeling and simulation and generating a subscriber-customized reduced-size feature set, the operations comprising:
acquiring, by a platform instance of the computing system, an input dataset comprising a set of structured items and unstructured textual data that relates to one or more items in the input dataset; and
reducing a size of the input dataset for transmission, via an electronic communication channel, to a target computing system by generating a labeled enriched feature set that corresponds to a set of automatically determined contextual items, comprising:
extracting the unstructured textual data from the input dataset;
using fuzzy logic, generating, by the platform instance, the set of contextual items using the extracted unstructured textual data;
labeling a set of items in the input dataset with the set of contextual items generated for a particular item in the set of items;
using a labeled particular item from the set of items in the labeled enriched feature set,
generating a set of entity behavior maps comprising one or more of:
a transaction map, a customer map, or a tradeline map,
wherein each entity behavior map in the set of entity behavior maps includes a set of features selected from the labeled enriched feature set based on a first particular label of a particular item in the set of items such that a particular entity behavior map corresponds to a particular contextual item;
for each of the set of entity behavior maps,
using a classification of the entity behavior map, the classification corresponding to the transaction map, the customer map, or the tradeline map, determining a set of operations to populate the entity behavior map, a particular operation in the set of operations being one of (i) AI/ML modeling operations performed by a trained neural network, (ii) an aggregation operation, or (iii) a summarization operation; and
using the particular contextual item that corresponds to the classification of the entity behavior map, performing operations comprising:
populating the entity behavior map;
using the classification and a target system identifier for the target computing system, querying a feature store to determine a particular dimension of the entity behavior map to transmit to the target computing system, wherein the particular dimension is a portion of the entity behavior map;
generating an electronic message for a reduced-size transmission, wherein the electronic message comprises the particular dimension; and
causing the target computing system to process the reduced-side transmission for publication of the dimension of the entity behavior map at the target computing system.

### Claim 2 (depends on 1)

The system of claim 1, wherein generating by the platform instance the set of contextual items comprises:
extracting a token from the unstructured textual data; and
referencing an ontology to determine whether the token matches an item in the ontology within a predetermined degree of certainty,
wherein the set of contextual items includes an item descriptor class from the ontology.

### Claim 3 (depends on 2)

The system of claim 2,
wherein the ontology corresponds to a feature definition criterion from the feature store associated with the target computing system, the target computing system being a particular subscriber computing system having a particular associated feature store; and
wherein a particular AI/ML modeling operation for the entity behavior map is performed based on the feature definition criterion.

### Claim 4 (depends on 3)

The system of claim 3,
wherein performing the AI/ML modeling operation comprises executing a machine learning model, and
wherein the machine learning model operates on items extracted from the input dataset,
the operations further comprising:
using the labeled particular item, generating an input feature according to the feature definition criterion corresponding to a particular label; and
using the input feature, executing a machine learning operation to perform the AI/ML modeling operation.

### Claim 5 (depends on 4)

The system of claim 4, the AI/ML modeling operation comprising at least one of item recurrence analytics, item tradeline analytics, item descriptor class analytics, item location analytics, entity income analytics, or entity employer analytics.

### Claim 6 (depends on 1)

The system of claim 1, the operations comprising acquiring the input dataset from the target computing system.

### Claim 7 (depends on 1)

The system of claim 1, the operations comprising acquiring the input dataset from a third-party computing system relative to the target computing system, wherein the input dataset includes data contributed by an entity associated with the target computing system.

### Claim 8 (depends on 1)

The system of claim 1, wherein the target computing system is one of a plurality of target computing systems and the platform instance is one of a plurality of platform instances, and wherein each particular target computing system of the plurality of target computing systems is associated with a particular platform instance configured for a particular subscriber.

### Claim 9 (depends on 1)

The system of claim 1, wherein labeling a particular item comprises generating metadata for the particular item using the contextual item.

### Claim 10

A computer-implemented method for artificial intelligence/machine learning (AI/ML) based modeling and simulation and generating a subscriber-customized reduced-size feature set, the method comprising:
acquiring, by a platform instance of a computing system, an input dataset comprising a set of structured items and unstructured textual data that relates to one or more items in the input dataset;
reducing a size of the input dataset for transmission, via an electronic communication channel, to a target computing system by generating a labeled enriched feature set that corresponds to a set of automatically determined contextual items, comprising:
extracting the unstructured textual data from the input dataset; and
using fuzzy logic, generating, by the platform instance, the set of contextual items using the extracted unstructured textual data;
labeling a set of items in the input dataset with the set of contextual items;
using a labeled particular item from the set of items in the labeled enriched feature set,
generating a set of entity behavior maps comprising one or more of: a transaction map, a customer map, or a tradeline map
wherein each entity behavior map in the set of entity behavior maps includes a set of features selected from the labeled enriched feature set based on a first particular label of the particular item such that a particular entity behavior map corresponds to a particular contextual item;
for each of the set of entity behavior maps,
using a classification of the entity behavior map, the classification corresponding to the transaction map, the customer map, or the tradeline map, determining a set of operations to populate the entity behavior map, a particular operation in the set of operations being one of (i) AI/ML modeling operations performed by a trained neural network, (ii) an aggregation operation, or (iii) a summarization operation; and
using the particular contextual item that corresponds to the classification of the entity behavior map, performing operations comprising:
populating the entity behavior map;
using the classification and a target system identifier for the target computing system, querying a feature store to determine a particular dimension of the entity behavior map to transmit to the target computing system, wherein the particular dimension is a portion of the entity behavior map;
generating an electronic message for a reduced-size transmission, wherein the electronic message comprises the particular dimension; and
causing the target computing system to process the reduced-side transmission for publication of the dimension of the entity behavior map at the target computing system.

### Claim 11 (depends on 10)

The method of claim 10, wherein generating by the platform instance the set of contextual items comprises:
extracting a token from the unstructured textual data; and
referencing an ontology to determine whether the token matches an item in the ontology within a predetermined degree of certainty,
wherein the set of contextual items includes an item descriptor class from the ontology.

### Claim 12 (depends on 11)

The method of claim 11,
wherein the ontology corresponds to a feature definition criterion from the feature store associated with the target computing system, the target computing system being a particular subscriber computing system having a particular associated feature store; and
wherein a particular AI/ML modeling operation for the entity behavior map is performed based on the feature definition criterion.

### Claim 13 (depends on 12)

The method of claim 12,
wherein performing the AI/ML modeling operation comprises executing a machine learning model, and
wherein the machine learning model operates on items extracted from the input dataset,
the operations further comprising:
using the labeled particular item, generating an input feature according to the feature definition criterion corresponding to a particular label; and
using the input feature, executing a machine learning operation to perform the AI/ML modeling operation.

### Claim 14 (depends on 13)

The method of claim 13, the AI/ML modeling operation comprising at least one of item recurrence analytics, item tradeline analytics, item descriptor class analytics, item location analytics, entity income analytics, or entity employer analytics.

### Claim 15 (depends on 10)

method of claim 10, further comprising acquiring the input dataset from the target computing system.

### Claim 16 (depends on 10)

The method of claim 10, further comprising acquiring the input dataset from a third-party computing system relative to the target computing system, wherein the input dataset includes data contributed by an entity associated with the target computing system.

### Claim 17 (depends on 10)

The method of claim 10, wherein the target computing system is one of a plurality of target computing systems and the platform instance is one of a plurality of platform instances, and wherein each particular target computing system of the plurality of target computing systems is associated with a particular platform instance configured for a particular subscriber.

### Claim 18 (depends on 10)

The method of claim 10, wherein labeling a particular item comprises generating metadata for the particular item using the contextual item.

### Claim 19

One or more non-transitory computer-readable media having computer-executable instructions stored thereon, the instructions, when executed by at least one process of a computing system, performing operations for artificial intelligence/machine learning (AI/ML) based modeling and simulation and generating a subscriber-customized reduced-size feature set, the operations comprising:
acquiring, by a platform instance of the computing system, an input dataset comprising a set of structured items and unstructured textual data that relates to one or more items in the input dataset; and
reducing a size of the input dataset for transmission, via an electronic communication channel, to a target computing system by generating a labeled enriched feature set that corresponds to a set of automatically determined contextual items, comprising:
extracting the unstructured textual data from the input dataset;
using fuzzy logic, generating, by the platform instance, the set of contextual items using the extracted unstructured textual data;
labeling a set of items in the input dataset with the set of contextual items generated for a particular item in the set of items;
using a labeled particular item from the set of items in the labeled enriched feature set,
generating a set of entity behavior maps comprising one or more of:
a transaction map, a customer map, or a tradeline map
wherein each entity behavior map in the set of entity behavior maps includes a set of features selected from the labeled enriched feature set based on a first particular label of a particular item in the set of items such that a particular entity behavior map corresponds to a particular contextual item;
for each of the set of entity behavior maps,
using a classification of the entity behavior map, the classification corresponding to the transaction map, the customer map, or the tradeline map, determining a set of operations to populate the entity behavior map, a particular operation in the set of operations being one of (i) AI/ML modeling operations performed by a trained neural network, (ii) an aggregation operation, or (iii) a summarization operation; and
using the particular contextual item that corresponds to the classification of the entity behavior map, performing operations comprising:
populating the entity behavior map;
using the classification and a target system identifier for the target computing system, querying a feature store to determine a particular dimension of the entity behavior map to transmit to the target computing system, wherein the particular dimension is a portion of the entity behavior map;
generating an electronic message for a reduced-size transmission, wherein the electronic message comprises the particular dimension; and
causing the target computing system to process a reduced-side transmission for publication of the dimension of the entity behavior map at the target computing system.

### Claim 20 (depends on 19)

The media of claim 19, wherein generating by the platform instance the set of contextual items comprises:
extracting a token from the unstructured textual data; and
referencing an ontology to determine whether the token matches an item in the ontology within a predetermined degree of certainty,
wherein the set of contextual items includes an item descriptor class from the ontology.

### Claim 21 (depends on 1)

The system of claim 1, the operations comprising:
reducing a size of at least a portion of the input dataset by quantizing values in the at least a portion of the input dataset; and
using the quantized values to generate numerical values for the particular dimension transmitted to the target computing system.
