ML-Based Sensitive Data Entry Prediction and Prevention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Customers and customer service representatives often accidentally enter sensitive information, such as personally identifiable information (PII), into incorrect form fields in electronic documents, leading to potential violations of regulations and damage to an organization's reputation due to unencrypted transmission and storage.
Innovation Solution
A system that uses non-biometric behavioral data to predict the likelihood of inappropriate entry of sensitive information into electronic form fields, providing user assistance through messages, chatbots, and preventing further input until incorrect data is removed, employing a machine learning model trained on past interactions to mitigate risks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If users are allowed to freely input data into electronic form fields, then ease of operation is improved, but the risk of inappropriate entry of sensitive information increases
Solution Approach 1:
The system performs preliminary actions by predicting the likelihood of inappropriate data entry before it occurs. The machine learning model analyzes user behavior patterns and form field characteristics in advance to identify high-risk scenarios, allowing the system to preemptively alert users or block potentially harmful data entry actions before sensitive information is compromised.
Solution Approach 2:
The system implements feedback mechanisms where users receive real-time alerts and guidance based on predicted risks. The machine learning model continuously monitors user interactions and provides feedback through notifications, tooltips, or blocking actions, creating a closed-loop system that adapts to user behavior and dynamically adjusts its protection strategies.
2Measurement precision
If a machine learning model is used to predict inappropriate data entry, then accuracy in detecting sensitive data risks is improved, but device complexity increases
Solution Approach 1:
The machine learning model performs self-service by automatically learning from user behavior patterns and form interaction data without requiring manual retraining for each new scenario. The system continuously adapts to evolving user behaviors and data entry patterns autonomously, reducing the need for complex manual configuration and model updates while maintaining high detection accuracy.
3Reliability
If user assistance messages and blocking actions are implemented, then reliability of sensitive data protection is improved, but loss of time increases
Solution Approach 1:
The system applies local quality by providing targeted, context-specific assistance only where needed. Rather than uniformly blocking all data entry actions, the machine learning model identifies specific form fields and user actions that pose high risks, applying protective measures locally only to those critical areas. This selective approach maintains high reliability for protected data while minimizing unnecessary time delays for legitimate data entry operations.
Data Source
AI summary
Disclosed aspects pertain to predicting a likelihood that a user enters sensitive data inappropriately and mitigating such a risk. An electronic form is monitored for user interaction. Non-biometric behavior data can be received or otherwise acquired for the user. A machine learning model can be invoked to determine a likelihood that the user will input sensitive data inappropriately into a form field based on the similarity of user non-biometric behavior data with historic non-biometric behavior data. User assistance, such as a message or warning, can be initiated when the likelihood satisfies a predetermined threshold to mitigate the risk of input of sensitive data inappropriately. The field can also be monitored to detect the presence of inappropriate sensitive data. The user can be prompted to redact or remove the sensitive data if detected.