Faasera Documentation

Faasera Profiler Policy Format Guide

This document describes the structure and components of a Faasera Policy Document, which governs how profiling, masking, and classification are performed across structured and un-structured datasets.

---

Overview

Faasera supports dynamic and declarative policy definitions using JSON. Policies can be used to:

• Profile sensitive data using built-in or custom recognizers
• Apply masking rules and functions
• Define exclusions and context-aware logic
• Integrate AI-powered hint services (like GLiNER)
• Manage encryption and masking configurations

---

Policy Structure

{
  "name": "default-policy",
  "config": {
    ...
  },
  "recognizers": {
    ...
  },
  "exclusion": {
    ...
  },
  "exclusionPatterns": [
    ...
  ],
  "crypto": {
    ...
  },
  "fpe": {
    ...
  },
  "graphical": {
    ...
  }
}

---

Recognizers

The recognizers section defines what types of data to detect.

"recognizers": {
"FIRSTNAME": true,
"EMAIL_ADDRESS": true,
"CREDITCARD_NUMBER": true
}

You can use internal types or enrich with AI-based hinting (classificationToHintService) for third-party models like GLiNER:

"classificationToHintService": {
"name": "FIRSTNAME",
"email": "EMAIL_ADDRESS"
}

---

Masking Rules

Defined in the maskConfig under config.

"config": {
"maskConfig": {
"FIRSTNAME": {
"type": "SEED",
"seedFileName": "firstnames.txt",
"seedSeparator": ","
},
"EMAIL_ADDRESS": {
"type": "PRESERVE",
"preserveStructure": true
}
}
}

Supported types include:

• SEED
• PRESERVE
• HASH
• REDACT
• RANDOM
• FPE (format-preserving encryption)

---

Exclusion

"exclusion": {
"global": ["test", "sample"],
"fileBased": {
"emails.csv": ["noreply@example.com"]
}
}

You can also use exclusionPatterns:

"exclusionPatterns": ["\d{4}-\d{4}"]

---

Crypto

"crypto": {
"algorithm": "AES",
"key": "base64-encoded-key=="
}

Optional fpe config for format-preserving encryption:

"fpe": {
"radix": 10,
"key": "hex-key",
"tweak": "tweak-key"
}

---

AI Hint Mapping

Use classificationToHintService in your policy when enriching labels from third-party models like GLiNER.

"classificationToHintService": {
"name": "FIRST_AND_LASTNAME",
"ssn": "SSN"
}

ℹ️ Note: Third-party models may be under separate licenses (e.g., GLiNER from HuggingFace). Refer to respective licenses for usage rights.

---

Format & Usage

Faasera supports both JSON and YAML formats.

• Use JSON for API integration
• Use YAML for human-readable editing
• Policy is stored as JSON in the internal metadata database

---

Sample Minimal Policy

{
  "name": "sample-policy",
  "recognizers": {
    "FIRSTNAME": true,
    "SSN": true
  },
  "config": {
    "maskConfig": {
      "FIRSTNAME": {
        "type": "SEED",
        "seedFileName": "firstnames.txt"
      },
      "SSN": {
        "type": "REDACT"
      }
    }
  }
}

---

📌 Notes

• dictionary-lookup and mask-functions are reserved for custom UDM usage and are not used in profiling policies directly.
• Make sure to test all custom policies before applying in production pipelines.
• You can upload and manage policy files via the Faasera UI or API.

---

For further examples, please visit faasera.ai or contact support.