Email Quarantine via Domain Relationship Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing solutions for blocking unwanted emails, such as gateway solutions and anti-spam software, are ineffective against previously unrecognized malware and targeted attacks, and rely on human error for mitigation.
Innovation Solution
A man-in-the-middle server agent that applies rules to identify suspicious emails and uses machine learning-based classification to quarantine or block emails by analyzing relationships between email domains, with optional sandbox testing and notification to recipients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If gateway solutions and anti-spam software are used to block unwanted emails, then recognized malware and spam can be filtered, but previously unrecognized malware and targeted attacks cannot be effectively blocked
Solution Approach 1:
The system performs preliminary analysis of email relationships between domains before emails arrive. By pre-establishing relationship graphs that map communication patterns between email domains, the system prepares detection mechanisms in advance, enabling it to quickly identify and block targeted attacks and new malware without relying on prior recognition of specific threats.
Solution Approach 2:
The invention introduces an intermediary relationship analysis layer between the email gateway and the recipient. This intermediary system analyzes the relational context between sending and receiving domains, acting as a mediator that determines whether to allow or block emails based on established relationship patterns, rather than relying solely on signature-based detection.
2Measurement precision
If relationship analysis between email domains is performed to improve detection accuracy, then targeted attacks can be identified, but system complexity increases
Solution Approach 1:
The system creates simplified copies or representations of complex email relationships through relationship graphs. Instead of analyzing all possible interactions between email domains in real-time, the system pre-processes and stores relationship patterns as graph structures, which can be quickly queried and compared during email filtering without requiring complex real-time computation.
3Reliability
If machine learning-based classification is used to quarantine emails, then unwanted emails are blocked effectively, but false positives may increase
Solution Approach 1:
The system incorporates feedback mechanisms where the outcomes of email classification decisions are fed back into the relationship graph analysis. When emails are quarantined or blocked, the system learns from these decisions by updating relationship patterns, allowing it to refine its classification accuracy over time and reduce false positives while maintaining effective blocking of unwanted emails.
Data Source
AI summary
A technique includes determining pairwise relationships among entities associated with a first electronic mail organization and entities associated with a second electronic mail organization. The technique includes controlling receipt of an electronic message originating from a sender associated with the first email organization based on the determined pairwise relationships.