Multi-Level Data Fusion System for Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Analysts in information security and fraud investigations face challenges in aggregating and analyzing data from various structured and unstructured sources to produce timely, accurate, and meaningful intelligence for situational awareness and decision-making.
Innovation Solution
An automated computer-implemented method and system for multi-level data fusion, which aggregates data from multiple sources, extracts features, enriches the data by categorizing it, and generates datasets to identify potentially fraudulent activity, using a graphical interface to propose actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If data is aggregated from multiple structured and unstructured sources, then the comprehensiveness of intelligence is improved, but the complexity of data processing increases
Solution Approach 1:
The system segments the complex data processing task into multiple distinct modules: data aggregation module, feature extraction module, data enrichment module, dataset generation module, and decision support module. Each module handles a specific aspect of the processing pipeline, making the overall complex system manageable and maintainable while comprehensively processing data from multiple sources
Solution Approach 2:
The patent introduces intermediate processing layers between raw data and final intelligence output. The feature extraction module serves as an intermediary that transforms raw aggregated data into structured features, and the data enrichment module further intermediates by adding contextual information. These intermediary layers simplify the processing of comprehensive data by breaking it down into manageable stages
2Measurement precision
If manual analysis of aggregated data is performed, then the accuracy of fraud detection is improved, but the time required for analysis increases
Solution Approach 1:
The system implements automated self-service capabilities where the data enrichment module automatically compiles data into categories, the dataset generation module automatically creates datasets for fraud identification, and the decision module automatically generates proposed actions. This automation maintains high accuracy previously achieved through manual analysis while dramatically reducing the time required
Solution Approach 2:
The patent replaces manual mechanical analysis processes with automated computational systems. The feature extraction module uses automated algorithms to identify patterns, the data enrichment module automatically categorizes information, and the decision module generates recommendations without human intervention, thereby reducing analysis time while maintaining or improving accuracy through consistent automated processing
3Reliability
If comprehensive data from multiple sources is collected, then the ability to identify fraudulent activity is improved, but the volume of data to be processed increases
Solution Approach 1:
The feature extraction module selectively extracts only the relevant features and information from the comprehensive aggregated data, separating useful signals from unnecessary volume. This extraction process maintains the reliability of fraud identification by focusing on critical features while reducing the overall data volume that needs to be processed in subsequent stages
Solution Approach 2:
The data enrichment module applies different processing qualities to different portions of the data based on their relevance and type. Critical fraud-related features receive more detailed enrichment and categorization, while less critical data receives minimal processing. This local quality approach ensures reliable fraud detection in high-priority areas while efficiently managing overall data volume
Data Source
AI summary
An embodiment of the present invention involves a computer implemented method and system for implementing data fusion comprising aggregating data from a plurality of sources via one or more computer networks, wherein the data comprises at least unstructured data; extracting one or more features from the aggregated data; enriching the extracted data by compiling the data into one or more categories; generating one or more datasets based on the enriched data for identifying potentially fraudulent activity; and identifying one or more proposed actions to address the potentially fraudulent activity using a graphical interface.