Phased Voice Call Fraud Detection Before Sensitive Data Disclosure
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to effectively detect fraudulent calls due to the dynamic nature of scammer tactics, including the use of recycled phone numbers and AI-generated voices, leading to increased inconvenience and potential financial and security risks for consumers.
Innovation Solution
A system that segments voice calls into discrete segments, analyzes each segment using machine learning models to identify patterns and intent, and provides real-time warnings to users before sensitive data disclosure, adapting to evolving fraudster tactics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing systems use static detection methods based on known spam numbers, then implementation simplicity is maintained, but detection reliability deteriorates due to recycled and AI-generated fraudulent calls
Solution Approach 1:
The call detection process is divided into multiple phases: initial call phase, intermediate phase, and critical phase. Each phase has specific analysis parameters and fraud score thresholds. This segmentation allows the system to adapt its detection strategy dynamically, improving reliability without requiring complete system redesign.
Solution Approach 2:
The system transitions from static number-based blocking to dynamic multi-phase analysis with evolving fraud scores. Fraud scores are updated continuously based on segment characteristics, allowing the system to adapt to new fraud tactics while maintaining manageable complexity through structured phase transitions.
2Measurement precision
If the system performs comprehensive real-time analysis of all call segments, then detection precision improves, but processing time increases
Solution Approach 1:
The system applies different levels of analysis intensity to different call phases. The initial call phase uses basic parameters, while the critical phase applies more intensive analysis only when fraud indicators are present. This partial action approach maintains high detection precision while avoiding unnecessary processing time for legitimate calls.
Solution Approach 2:
By dividing the call into discrete segments with specific analysis parameters for each phase, the system processes only relevant features at each stage rather than analyzing all possible parameters throughout the entire call, reducing overall processing time while maintaining precision.
3Object-affected harmful factors
If the system provides warnings during ongoing calls, then user protection improves, but user convenience deteriorates due to potential false alarms
Solution Approach 1:
The system provides warnings in advance during the critical phase, before the scammer can obtain sensitive information. This preliminary anti-action allows users to take protective measures while minimizing disruption to legitimate calls, balancing protection with convenience through phase-based threshold evaluation.
4Adaptability or versatility
If the system uses weighted fraud scores that account for previous segments, then detection adaptability improves, but computational complexity increases
Solution Approach 1:
The weighted fraud score mechanism incorporates feedback from previous segments into current analysis. Each segment's fraud score is calculated considering historical patterns, allowing the system to adapt to evolving fraud tactics. This feedback loop provides high adaptability while maintaining reasonable computational complexity through efficient score updating.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system and method for detecting fraudulent call activity include segmenting an ongoing voice call between a user and a second party into discrete segments while the call is in progress. The method analyzes respective discrete segments and assigning per-segment weighted fraud scores, where each weighted fraud score accounts for the weighted fraud score of a previous segment. Based on these per-segment weighted fraud scores, the method determines that the voice call is likely a fraudulent call. After making this determination, the method provides a human-perceptible warning to the user before the user discloses sensitive user data.