Ordered Rule Sets for Fraud Detection in Electronic Transactions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing fraud detection systems for electronic transactions, particularly in blockchain transactions, face inefficiencies and high false positive rates, making them ineffective for accurately identifying fraudulent activities across various transaction types.
Innovation Solution
A processing system generates ordered fraud and approval rules using a decision tree method, applying multiple rules to transaction data to filter out fraudulent transactions and ordering them based on effectiveness, which can be applied to both traditional and blockchain transactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional fraud scoring algorithms are applied to analyze trends of past transactions, then fraud scores can be generated for new transactions, but the process is inefficient and ineffective, especially for blockchain transactions
Solution Approach 1:
The patent segments the fraud detection process into multiple independent rules that can be applied sequentially. Each rule targets specific fraud patterns (e.g., rule 101 for blockchain transactions, rule 102 for traditional transactions), allowing the system to divide the complex detection task into manageable parts that can be processed efficiently and independently
Solution Approach 2:
The system performs preliminary actions by pre-defining multiple fraud detection rules based on historical data and patterns before actual fraud detection is needed. These rules are prepared in advance and can be directly applied to new transactions without requiring real-time analysis of past transaction trends, significantly improving efficiency
2Adaptability or versatility
If multiple fraud factors are applied concurrently to traditional electronic payment transactions, then fraud scoring can be performed, but the system generates high false positive rates and lacks effectiveness for blockchain transactions
Solution Approach 1:
The patent applies local quality by tailoring specific fraud detection rules to specific transaction types. Different rules are designed for blockchain transactions (e.g., checking for unconfirmed transactions, unusual network activity) versus traditional electronic payment transactions (e.g., checking for known fraud patterns, transaction velocity). This localized approach ensures each rule type is optimized for its specific context, reducing false positives while maintaining versatility across transaction types
Solution Approach 2:
The system achieves universality by creating a unified fraud detection framework that can handle multiple transaction types through a common rule-based architecture. The same fundamental approach of applying predefined rules can be used for both blockchain and traditional transactions, with the specific rules adapted to each type's characteristics, making the system universally applicable without requiring separate systems for each transaction type
Data Source
AI summary
A method for generating fraud and approval rules for a decision engine for detecting fraud in electronic transactions includes: receiving transaction data for a plurality of electronic transactions, the transaction data including a fraud determination and data values for the respective electronic transaction; applying a first rule of a plurality of rules to the electronic transactions to identify a first subset of transactions that satisfy the first rule and include a fraud determination indicative of fraud; filtering the first subset of transactions out of the plurality of electronic transactions; repeating the application step and filtering step using additional rules of the plurality of rules until a threshold criteria is met; and generating a rule order for the first rule and the additional rules based on at least a size of the subset of transactions identified using the respective rule.