Fuzzy Pattern Matching for Real-Time User Account Fraud Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting patterns in new user accounts in service systems are inadequate in real-time environments, as malicious parties have developed strategies to avoid direct linking between accounts, making it difficult to detect fraudulent or malicious activities effectively.
Innovation Solution
Implementing fuzzy pattern matching using machine learning algorithms that match accounts based on recognition of patterns in information such as email addresses and IP geographic locations, allowing new user accounts to be linked to predefined 'fuzzy' nodes in a graphical database, enabling real-time detection of potential account issues.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If direct linking methods are used to detect fraudulent accounts, then the detection method is simple and fast, but malicious parties can easily avoid detection by avoiding direct linking between accounts
Solution Approach 1:
The patent transforms the detection approach from exact matching of account properties to fuzzy pattern matching by changing the parameters used for comparison. Instead of requiring direct links between accounts, the system uses pattern similarity thresholds to identify potentially fraudulent accounts, thereby improving detection reliability while maintaining manageable system complexity
Solution Approach 2:
The patent replaces the mechanical direct-linking detection method with a machine learning-based fuzzy pattern matching system. This substitution enables the system to detect complex patterns and relationships that go beyond simple direct links, improving fraud detection accuracy without requiring overly complex manual analysis
2Reliability
If fuzzy pattern matching is implemented to improve fraud detection, then detection accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The patent implements preliminary action by pre-computing and storing pattern representations in a graph database before actual account verification is needed. This allows the system to quickly query and compare patterns during real-time account creation, improving detection accuracy while minimizing processing time and computational resource usage during critical verification moments
Data Source
AI summary
Various techniques are disclosed for providing dynamic, real-time pattern detection and linking between newly created user accounts and existing user accounts. Certain solutions include assessing matches of a new user account to nodes in a graphical representation of a machine learning algorithm based on predefined patterns in the properties of the new user account. The nodes in the graphical representation may each have different predefined patterns that have been determined based on patterns in previous information for user accounts, which can also be updated in real-time as needed. Accordingly, when a new user account is matched (e.g., assigned) to a node that has accounts with known issues associated with the node, the new user account may be flagged for increased scrutiny or other solutions.