Watchlist Identity Resolution Using Graph-Based Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional watchlist identification methods rely heavily on manual comparison of ever-changing lists, leading to errors due to name misspellings, incorrect PII, and lack of consideration for aliases, which compromises the integrity of transactions by introducing false positives and negatives.
Innovation Solution
A machine learning system and method that utilizes unsupervised semantic identity modeling and graph-based clustering to build holistic identity profiles, incorporating personal and network-ascertained contextual data, and applies machine learning models to determine watchlist candidacy with high accuracy and precision, using techniques like NLP and LSTM algorithms to correct errors and derive watchlist tags.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual review of watchlists is used, then human judgment and contextual understanding are applied, but the process is time-consuming and error-prone due to the overwhelming volume of data and ever-changing listings
Solution Approach 1:
The patent replaces manual mechanical review processes with automated machine learning systems that use NLP and graph-based clustering algorithms to process watchlist data, eliminating the need for human reviewers to manually compare enormous amounts of changing listings while maintaining high accuracy through contextual inspection and semantic matching
Solution Approach 2:
The system performs self-improving automated review by continuously learning from feedback loops where human reviewers validate model predictions, allowing the system to automatically refine its algorithms and improve accuracy over time without requiring continuous manual intervention
2Measurement precision
If manual comparison of watchlists is performed, then contextual understanding is applied, but errors are introduced due to name misspellings, incorrect PII, and lack of consideration for aliases
Solution Approach 1:
The patent replaces manual mechanical comparison with automated NLP-based semantic matching that can recognize name variations, misspellings, and aliases by analyzing contextual relationships and graph-based connections between entities, thereby improving precision while maintaining reliability through multi-factor verification
Solution Approach 2:
The system moves beyond simple string-matching to multi-dimensional analysis by incorporating graph-based clustering that examines relationships between entities across multiple data dimensions including contextual information, network connections, and semantic similarities, enabling detection of matches that would be missed by traditional manual review
3Measurement precision
If extensive feature engineering is performed to improve accuracy, then the system becomes more complex, but the computational resources and processing time required increase significantly
Solution Approach 1:
The patent performs feature engineering and model development in advance during the training phase, pre-processing and transforming data into meaningful features that can be quickly applied during production operations, thereby achieving high accuracy without requiring complex real-time processing during actual watchlist comparisons
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided are a method and system for identity correlation between a transaction applicant (TA) and a watchlist entity (WE). Preexisting watchlist data and other aggregated identity data (AID) are processed to provide for comparison to a collective identity of at least the TA. Using various categorizations for the AID and the collective identity, watchlist tags are generated that can then be matched to the collective identity. As a result of the matching, a watchlist candidacy demonstrating a probability that the identity of the TA does or does not correspond to that of the WE can be generated.