Data Classification Heuristic Tool

Classifies data sensitivity using pattern-based heuristics to detect PII, financial data, health data, and other sensitive information.

Installation

npm install @tpmjs/tools-data-classification-heuristic

Usage

import { dataClassificationHeuristic } from '@tpmjs/tools-data-classification-heuristic';

const result = await dataClassificationHeuristic.execute({
  text: "Contact John Doe at john.doe@example.com or call 555-123-4567. SSN: 123-45-6789"
});

console.log(result);
// {
//   classification: 'restricted',
//   signals: [
//     { type: 'Email', severity: 'medium', description: 'Email address detected', matches: 1 },
//     { type: 'Phone', severity: 'medium', description: 'Phone number detected', matches: 1 },
//     { type: 'SSN', severity: 'critical', description: 'Social Security Number detected', matches: 1 }
//   ],
//   confidence: 0.8,
//   summary: {
//     totalSignals: 3,
//     highestSeverity: 'critical',
//     categories: ['PII']
//   }
// }

Classification Levels

• public - No sensitive data detected, safe for public distribution
• internal - Low-medium sensitivity data, internal use only
• confidential - High sensitivity data, restricted distribution
• restricted - Critical data (SSN, credentials, financial), highly restricted

Detected Patterns

PII (Personal Identifiable Information)

• Social Security Numbers (SSN)
• Email addresses
• Phone numbers
• Dates of birth
• Physical addresses

Financial Data

• Credit card numbers
• Bank account numbers
• Routing numbers
• Salary information

Health Data (HIPAA)

• Medical record numbers
• Diagnoses
• Prescriptions

Government IDs

• Passport numbers
• Driver license numbers

Authentication

• API keys
• Passwords
• Access tokens

Technical

• IP addresses

Output Schema

interface DataClassification {
  classification: 'public' | 'internal' | 'confidential' | 'restricted';
  signals: Array<{
    type: string;
    pattern: string;
    severity: 'low' | 'medium' | 'high' | 'critical';
    description: string;
    matches?: number;
  }>;
  confidence: number; // 0-1 scale
  summary: {
    totalSignals: number;
    highestSeverity: string;
    categories: string[];
  };
}

Use Cases

• Data Loss Prevention (DLP) - Scan documents before sharing
• Compliance Auditing - Identify sensitive data in databases
• Email Filtering - Classify email content sensitivity
• Document Review - Automatically classify documents for access control
• Privacy Impact Assessment - Detect PII in data processing activities

Limitations

• Heuristic-based detection (pattern matching only)
• May produce false positives (e.g., random number sequences)
• Does not understand context or semantic meaning
• Should be used as a first-pass filter, not definitive classification
• Cannot detect all types of sensitive data (e.g., trade secrets require domain knowledge)

License

MIT