Systems and methods for identifying electronic accounts based on near real time machine learning in an electronic network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Accurate, efficient, and near real-time scoring of electronic accounts' propensity to perform tasks, such as cybersecurity threats, is challenging due to data being spread across multiple applications with different formats and storage mechanisms, leading to unstructured datasets that are difficult to parse and understand.
Innovation Solution
A system utilizing a trained machine learning model that standardizes data from multiple applications, generates a unique entity identifier, and calculates propensity scores based on entity-specific requirements, optimizing resource usage and improving accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If data from multiple applications is collected and processed to determine electronic account propensity scores, then scoring accuracy is improved, but system complexity increases due to different data formats and storage mechanisms
Solution Approach 1:
The patent introduces a central processing system that acts as an intermediary between multiple applications with different data formats and the propensity scoring function. This intermediary collects data from various applications, standardizes their formats, and processes them through a unified machine learning model, thereby resolving the complexity issue while maintaining scoring accuracy.
Solution Approach 2:
The system creates a universal data processing framework that can handle multiple data formats and storage mechanisms from different applications through a single standardized interface. The machine learning model serves as a universal scorer that can evaluate diverse data types, enabling one system to perform multiple scoring functions across different applications.
2Speed
If near real-time scoring is implemented across multiple applications, then response speed is improved, but computing resource consumption increases
Solution Approach 1:
The patent segments the scoring process into modular components: data collection from individual applications, data standardization, and propensity scoring through a machine learning model. This segmentation allows the system to process data from multiple applications efficiently by handling each segment independently while maintaining near real-time performance.
Solution Approach 2:
The system performs preliminary data collection and standardization actions in advance, preparing data in a standardized format before it reaches the scoring stage. This preliminary processing enables faster scoring operations by eliminating the need for format conversion during real-time scoring, thus reducing computing resource consumption while maintaining speed.
3Ease of operation
If unstructured datasets from multiple applications are parsed and standardized, then data usability is improved, but processing time increases
Solution Approach 1:
The patent introduces a data standardization layer as an intermediary between the unstructured datasets from multiple applications and the propensity scoring function. This intermediary automatically parses and transforms diverse data formats into a unified structured format, making the data usable while managing processing time through efficient transformation algorithms.
Data Source
AI summary
Systems, computer program products, and methods are described herein for systems and methods for identifying electronic accounts based on near real time machine learning in an electronic network. The present disclosure is configured to: identify at least one data storage component associated with an at least one application; determine at least one current requirement of the data; receive at least one data segment from the data storage component; collect at least one current element from at least one data segment; reformat and aggregate the at least one current element and link the at least one current element to a unique entity identifier; apply a trained machine learning model to the at least one current element; generate a new data segment; and generate a propensity score based on at least one of unique entity identifier, at least one current element, or at least one new data segment.