Multi-factor Scam Call Detection Using Graph and Entropy Analytics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current caller ID and scam alerting systems provide insufficient context for users to differentiate between wanted and unwanted calls, as they lack comprehensive information about caller reliability and behavior, and can be easily bypassed by scammers through number spoofing.
Innovation Solution
Implementing a multi-factor scam call detection and alerting system that analyzes call record data using graph analytics and entropy analysis to generate reliability and behavior scores, which are then displayed to users as part of incoming call notifications, along with feedback mechanisms to refine these scores and identify potential scams.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a scam telephone number database is used to identify suspicious calls, then scam calls can be detected and warned to users, but the system cannot identify all suspicious calls because the database does not have a complete list of scammer numbers
Solution Approach 1:
The patent segments the scam detection task into multiple independent analysis components: graph analytics to build caller relationship networks, entropy analysis to detect abnormal calling patterns, and traditional database matching. This segmentation allows each component to specialize in different aspects of scam detection, improving overall reliability without requiring a complete scammer number database.
Solution Approach 2:
The system implements multi-functionality by combining multiple detection methods (graph analytics, entropy analysis, database matching) into a unified scam detection framework. This allows the system to handle diverse scamming techniques and identify suspicious calls even when the caller number is not in the database, making the detection mechanism more universal and robust.
2Ease of operation
If scammers spoof telephone numbers of innocent third-parties, then they can bypass the scam alerting function, but this creates insufficient context for users to differentiate between wanted and unwanted calls
Solution Approach 1:
The system uses feedback mechanisms where call outcomes (answered, rejected, reported as scam) are fed back into the graph analytics and entropy analysis models. This continuous feedback loop allows the system to learn from user behavior and improve its ability to provide accurate context information, helping users differentiate between wanted and unwanted calls even when number spoofing occurs.
Solution Approach 2:
The patent adds another dimension to caller identification by moving beyond traditional number-based matching to multi-dimensional analysis including graph-based relationship metrics, entropy-based behavioral patterns, and network connectivity analysis. This dimensional expansion provides rich context information that helps users distinguish legitimate calls from spoofed scam calls.
3Loss of information
If traditional caller ID information is provided, then users can see telephone number and pre-registered name, but this information is insufficient for users to make informed decisions about answering calls
Solution Approach 1:
The system creates a composite information profile for each caller by combining multiple data sources: traditional caller ID information, graph analytics results showing relationship strength, entropy analysis indicating behavioral normality, and network connectivity patterns. This composite information structure provides users with comprehensive context to make informed calling decisions.
Solution Approach 2:
The patent transforms static caller ID information into dynamic, multi-parameter assessments by calculating reliability scores and behavior scores that change based on observed calling patterns and network relationships. These parameter changes provide users with adaptive information that reflects the current trustworthiness of each caller.
Data Source
AI summary
Call record data for telephone numbers of callers are received. A telephone number of a caller is determined as used to place a telephone call to a recipient telephone number of a call recipient. The call record data are analyzed using graph analytics to generate a reliability score for the telephone number that represents a likelihood that the caller is known to the call recipient. The call record data are analyzed using entropy analysis to generate a behavior score for the telephone number that represents a behavior trustworthiness of the caller. The reliability score, the behavior score, and the telephone number of the caller are sent to a user device of the call recipient for the user device to generate a first indication that represents a value of the reliability score and a second indication that represents a value of the behavior score for display along with the telephone number.