Incoming Communication Spam Screening Using Profile Metadata
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately distinguish between genuine and spam communications, particularly in the context of evolving cyber threats and the proliferation of generative artificial intelligence, posing risks to both individuals and organizations.
Innovation Solution
A system utilizing Large Language Models (LLMs) and Generative Artificial Intelligence (GenAI) integrated into wireless telecommunication networks and user equipment (UEs) to analyze metadata and communication patterns, enhancing the ability to categorize and tag messages as spam or valid, leveraging databases like CNAM and ENAM to provide contextual information and user-specific data for improved decision-making.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional spam filtering methods are used, then device complexity remains low, but measurement precision of spam detection deteriorates due to sophisticated hacking techniques
Solution Approach 1:
The patent introduces an intermediary AI model that acts as a mediator between incoming communications and the user. This model analyzes communication patterns, metadata, and content to determine spam likelihood, thereby improving detection accuracy without requiring the user's device to implement complex detection algorithms directly. The intermediary processing layer handles the computational complexity while providing simplified spam indicators to users.
Solution Approach 2:
The patent replaces traditional mechanical spam filtering methods (keyword matching, blacklist/whitelist systems) with AI-based analytical models. These models use machine learning to identify spam patterns, analyze communication behavior, and assess risk levels, thereby achieving superior detection precision without relying on rigid rule-based systems that are easily bypassed by sophisticated hackers.
2Measurement precision
If AI models are deployed on user equipment, then spam detection accuracy improves, but use of energy increases due to computational requirements
Solution Approach 1:
The patent segments the spam detection system into multiple components: lightweight on-device AI models for initial assessment, cloud-based analytics for comprehensive analysis, and collaborative networking for shared threat intelligence. This segmentation allows energy-intensive operations to be distributed to cloud infrastructure while maintaining accurate local detection capabilities with minimal energy consumption.
Solution Approach 2:
The patent implements partial AI processing on user devices, applying AI models selectively based on communication risk indicators rather than analyzing every message in full detail. The system performs preliminary screening with minimal computational resources and only activates full AI analysis for suspicious communications, thereby achieving high detection accuracy while minimizing overall energy consumption.
3Measurement precision
If comprehensive metadata analysis is performed, then spam detection accuracy improves, but loss of time increases due to extensive processing requirements
Solution Approach 1:
The patent performs preliminary analysis of communication metadata (sender information, communication patterns, device profiles) before the actual communication occurs. By pre-establishing baseline profiles and detecting anomalies in advance, the system can quickly assess spam likelihood without requiring extensive real-time processing of communication content, thereby maintaining high accuracy while reducing processing delays.
Solution Approach 2:
The patent implements feedback mechanisms where detection results, user confirmations, and communication outcomes are continuously fed back into the AI models. This feedback loop enables the system to learn from past decisions, refine its analysis priorities, and progressively improve detection accuracy over time while reducing the computational burden required for each individual assessment through experience-based optimization.
Data Source
AI summary
The system receives a request for communication from an originator UE. The request for communication includes a unique identifier of the originator UE and a unique identifier of a receiver UE. The system obtains profile information of the originator UE including a name or a region of the originator UE. The system obtains profile information of the receiver UE including a communication history or calendar entry of the receiver UE. Based on the profile information of the originator UE and the receiver UE, the system determines whether the communication is spam. If the communication is valid, the system routes the communication to the originator UE. If the communication is spam, the system indicates to the receiver UE that there is an incoming communication that is likely spam. The system stores in a database the unique identifier of the originator UE and the determination of whether the communication is spam.