Email Sensitive Data Detection Using Trained Evaluation Function
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing electronic messaging technologies lack effective security measures to prevent the dissemination of sensitive data, such as passwords, which can be compromised by malicious actors, leading to unauthorized access to protected accounts and assets.
Innovation Solution
A computer-implemented method that analyzes email messages using a trained evaluation function, leveraging a character lookup table and machine learning models to identify sensitive data tokens, quarantines messages containing passwords, and informs users to remove them before dissemination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If electronic messaging technologies are used to share information, then communication efficiency and information sharing capability are improved, but security risks and the likelihood of sensitive data leakage increase
Solution Approach 1:
The system performs preliminary scanning and analysis of email messages before they are sent or received, identifying sensitive data patterns in advance. The trained evaluation function pre-processes incoming messages to detect passwords and other sensitive information, allowing preventive actions to be taken before the data can be compromised
Solution Approach 2:
The patent introduces an intermediary security system that acts as a mediator between email senders and recipients. This intermediary component intercepts, analyzes, and evaluates email content using trained models, and can quarantine or block messages containing sensitive data, thereby protecting the communication channel without preventing legitimate communication
2Reliability
If sensitive data is detected and blocked in email messages, then security and prevention of data breaches are improved, but communication flow and message delivery may be disrupted
Solution Approach 1:
The system implements feedback mechanisms where the trained evaluation function continuously learns from detected patterns and adjusts its detection criteria. The system provides feedback to users about blocked messages, allowing them to appeal or correct false positives, thereby maintaining security while reducing unnecessary disruptions to legitimate communication
Solution Approach 2:
The patent applies different levels of scrutiny and detection sensitivity to different parts of the email system. Instead of uniformly blocking all messages with detected patterns, the system selectively applies security measures based on the specific context, message type, and detected sensitivity level, allowing legitimate communications to flow while blocking only genuinely risky messages
3Measurement precision
If advanced machine learning models are used to detect sensitive data, then detection accuracy and sensitivity are improved, but system complexity and computational resources increase
Solution Approach 1:
The patent segments the email analysis process into distinct stages: initial filtering using simple pattern matching, followed by more sophisticated analysis using trained evaluation functions for messages that pass the first filter. This segmented approach allows high accuracy for critical detection while keeping overall system complexity manageable by applying complex models only where necessary
Solution Approach 2:
The system replaces manual security review processes with automated machine learning models. The trained evaluation functions automatically analyze email content, identify sensitive data patterns, and make blocking decisions without human intervention, thereby achieving high detection accuracy while reducing the operational complexity that would otherwise require extensive human security teams
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The described technologies leverage a trained evaluation function to analyze an email message to determine if a password is included in the text of the email message. The text of the email message may be vectorized using a character lookup table including vector values for each ASCII character. The trained evaluation function analyzes the vectorized text to determine if a password is included in the text of the mail message. An email message found to include a password may be placed in a quarantine storage to at least temporality prevent the email message from being disseminated to a recipient.