Anomalous User Activity Detection via Transaction Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional fraud detection methods in ERP systems are labor-intensive and often fail to timely detect complex fraudulent activities, leading to significant financial losses due to delayed detection of suspicious user behavior.
Innovation Solution
An automated system that processes transaction data to identify anomalous user activity by grouping transactions, generating user and transaction profiles, and determining relationships between them, flagging activities that deviate from predetermined parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If automated transaction screening is implemented, then detection speed and timeliness improve, but system complexity increases
Solution Approach 1:
The system segments the complex fraud detection task into distinct functional modules: transaction data reception module, grouping module (that clusters transactions by type, user, and time), anomaly detection module (that applies predetermined parameters), and flagging module. This segmentation allows each module to handle specific aspects independently, improving processing speed while managing complexity through modular design.
Solution Approach 2:
The patent introduces an intermediary anomaly detection mechanism that acts as a mediator between raw transaction data and auditor review. The system uses predetermined parameters and grouping logic as intermediaries to automatically filter and prioritize transactions, enabling fast automated screening without requiring auditors to directly analyze all raw data, thus balancing speed with manageable system complexity.
2Measurement precision
If manual fraud detection audits are performed, then detection accuracy can be maintained, but time consumption and resource requirements increase
Solution Approach 1:
The system performs preliminary actions by automatically grouping transactions by type, user, and time period before auditor review. Predetermined parameters are established in advance to define what constitutes anomalous behavior. This preliminary processing filters out normal transactions, presenting only suspicious patterns to auditors, thereby maintaining detection accuracy while dramatically reducing the time and resources required for manual review.
Solution Approach 2:
The system incorporates feedback mechanisms where flagged anomalous transactions are reviewed by auditors, and their decisions feed back into refining the predetermined parameters and grouping logic. This continuous feedback loop allows the system to learn from actual cases, improving detection accuracy over time while maintaining efficient automated processing speeds.
3Reliability
If comprehensive transaction analysis is conducted, then fraud detection capability improves, but processing time increases
Solution Approach 1:
The comprehensive transaction analysis is segmented into hierarchical grouping levels: first by transaction type, then by user, and finally by time period. This multi-level segmentation allows the system to analyze transactions comprehensively at each level without processing the entire dataset monolithically, improving fraud detection capability through thorough analysis while reducing overall processing duration through divide-and-conquer strategy.
Solution Approach 2:
The system applies partial action by focusing analysis only on transactions that meet predetermined anomaly criteria after grouping, rather than analyzing every transaction in detail. The grouping process identifies suspicious patterns, and only those flagged transactions receive comprehensive analysis, enabling reliable fraud detection with reduced processing time by avoiding excessive analysis of normal transactions.
Data Source
AI summary
Systems, methods and articles for determining anomalous user activity are disclosed. Data representing a transaction activity corresponding to a plurality of user transactions can be received and user transactions can be grouped according to types of user transactions. The transaction activity can be determined to be anomalous in relation to the grouped user transactions based on a predetermined parameter.