Probabilistic Multi-Attribute Linking for Scalable Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing online transaction platforms face challenges in identifying fraudulent activities involving multiple accounts due to the rigidity of deterministic entity resolution techniques, which often miss instances where accounts should be linked, exposing them to liability for non-compliance with anti-money laundering requirements, and subjective approaches are not scalable or certain.
Innovation Solution
A probabilistic account linking approach is employed, defining linking strategies based on account attributes, generating average linking probabilities using account data, and applying these probabilities to determine the likelihood that two accounts belong to the same entity, allowing for scalable and certain account linking.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If deterministic entity resolution techniques are used to link accounts, then the system provides clear and certain account linking decisions, but the system misses instances where accounts should be linked, reducing fraud detection effectiveness
Solution Approach 1:
The patent transforms the account linking problem from a deterministic binary decision to a probabilistic multi-parameter assessment. Multiple account attributes (email, phone, IP address, device identifiers, behavioral patterns) are evaluated with assigned weights to generate a similarity score, enabling the system to capture nuanced relationships that deterministic rules miss while maintaining decision certainty through threshold-based classification
Solution Approach 2:
The patent creates a composite linking decision by combining multiple independent account attributes and their respective matching scores into a unified similarity assessment. This composite approach integrates diverse data sources (personal information, device data, transaction patterns) to form a comprehensive view of account relationships, improving both accuracy and fraud detection coverage
2Adaptability or versatility
If subjective approaches are used to link accounts, then the system can identify nuanced account relationships, but the approach is not scalable and introduces uncertainty
Solution Approach 1:
The patent implements a self-service system where the probabilistic model automatically evaluates account relationships without requiring manual expert review. The system independently computes similarity scores, applies threshold rules, and generates linking decisions at scale, eliminating the bottleneck of subjective human assessment while preserving the ability to identify nuanced relationships through learned patterns
Solution Approach 2:
The patent replaces the mechanical subjective judgment process with an automated computational system. Instead of relying on human experts to manually assess account relationships, the system uses algorithmic probability calculations and similarity scoring to objectively evaluate and compare account pairs, enabling scalable processing while maintaining consistent decision criteria
3Reliability
If multiple account attributes are analyzed using probabilistic methods, then the system improves fraud detection capability, but the complexity of the linking system increases
Solution Approach 1:
The patent segments the complex account linking problem into distinct, manageable components: attribute extraction (separating different account data types), similarity computation (calculating matching scores for each attribute pair), and decision-making (applying thresholds to generate linking decisions). This segmentation allows each component to be optimized independently and simplifies the overall system architecture despite handling multiple attributes
Solution Approach 2:
The patent manages complexity by transforming multiple account attributes into a single unified similarity score through weighted aggregation. This parameter transformation consolidates diverse data dimensions (email matches, phone matches, device similarities) into one comprehensible metric that feeds into the decision rule, simplifying the complexity of multi-attribute analysis while preserving fraud detection capability
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Some aspects relate to technologies for probabilistic account linking, for instance, to perform fraud detection on online transaction platforms. In accordance with some configurations, linking strategies are defined for linking accounts based on account attributes. An average linking probability is generated for each linking strategy using account data for accounts on an online transaction platform, and the average linking probabilities are stored. To determine whether to link two accounts, linking strategies shared by the two accounts are identified, an account linking probability for the two accounts is generated using the average linking probabilities for the linking strategies shared by the two accounts, and the account linking probability is compared against a threshold. If the account linking probability satisfies the threshold, the accounts are linked and an action is taken based on the account linking.