Machine Learning Source Allocation for Exchange-Level Records

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current systems face challenges in accurately allocating resource portions from multiple sources to items associated with records due to insufficient item-level data, leading to incorrect allocations and inefficient resource utilization.

Innovation Solution

A machine learning-based system that retrieves a rule set and analyzes exchange-level data to determine item-level data, allowing for precise source association and allocation, thereby modifying records to indicate the correct sources and storing this information for improved resource management.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Use of energy by moving object

If exchange-level data is used for source allocation, then processing resources are conserved, but allocation accuracy deteriorates

Engineering Contradiction:

Improveprocessing resourcesVSAvoidallocation accuracy

Core Design Contradiction:

Use of energy by moving objectVSMeasurement precision

Solution Approach 1:

The patent segments the data into exchange-level data (for processing efficiency) and item-level data (for allocation accuracy). The machine learning model segments the analysis by predicting item-level characteristics from exchange-level data, allowing the system to process at the coarser exchange level while still achieving fine-grained allocation accuracy through the ML predictions.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If item-level data is obtained through machine learning analysis, then allocation precision is improved, but processing time increases

Engineering Contradiction:

Improveallocation precisionVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The machine learning model is trained in advance on historical data to learn the relationships between exchange-level and item-level characteristics. During actual source allocation, the pre-trained model quickly predicts item-level data from exchange-level data, avoiding the need for time-consuming real-time analysis of detailed item information while still achieving accurate allocation.

Inventive Principle:

Principle #10Preliminary action

3Loss of energy

If machine learning model analyzes exchange-level data, then network resources are conserved, but data granularity deteriorates

Engineering Contradiction:

Improvenetwork resourcesVSAvoiddata granularity

Core Design Contradiction:

Loss of energyVSLoss of information

Solution Approach 1:

The machine learning model acts as an intermediary that translates exchange-level data into predicted item-level characteristics. Instead of retrieving actual item-level data from databases (which would consume network resources), the ML model generates synthetic item-level information that preserves the necessary granularity for accurate source allocation while avoiding network overhead.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20240412102A1Machine learning analysis of a record

Publication Date: 2024.12.12 CAPITAL ONE SERVICES LLC

AI summary

In some implementations, a device may retrieve a rule set, associated with an account, indicating one or more rules for associating a source, of a first source and a second source associated with the account, with an item based on item-level data of the item. The device may retrieve information, associated with the account, indicating one or more records that indicate first data at an exchange level of data. The device may analyze the first data associated with a record, of the one or more records, to determine item-level data for respective items of one or more items associated with the record. The device may determine, based on the rule set and the item-level data, sources associated with respective items of the one or more items.