README.md

metadata

language:
  - fa
license: other
multilingual: false
pretty_name: encrypted_legal_dataset
task_categories:
  - other
task_ids: []
source_datasets: []
dataset_info:
  features:
    - name: AnonymizedJudge_text
      dtype: string
    - name: AnonymizedJSS_text
      dtype: string
    - name: UniqueTables
      dtype: string
    - name: UniqueItemArray
      dtype: string
    - name: JSSType
      dtype: string
    - name: othersdoc
      dtype: string
    - name: legal_references
      dtype: string
  splits:
    - name: train
      num_bytes: 275010760
      num_examples: 9404
  download_size: 275143859
  dataset_size: 275010760
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

ara_v7

ara_v7 is a dataset where the text column has been encrypted with AES-GCM (AES-256) to preserve privacy while still allowing distribution.

Dataset Description

• Columns:
  • score: floating-point metadata value
  • text: Base64-encoded string containing AES-GCM encrypted text
• Encryption:
  • AES-256 in GCM mode

Usage

You can load the dataset using the 🤗 Datasets library:

import base64
from cryptography.hazmat.primitives.ciphers.aead import AESGCM
from datasets import load_dataset

# 🔑 Replace this with the Base64 key provided securely
key_b64 = "PASTE-YOUR-KEY-HERE"
key = base64.b64decode(key_b64)
aesgcm = AESGCM(key)

def decrypt(token: str) -> str:
    data = base64.b64decode(token.encode())
    nonce, ciphertext = data[:12], data[12:]
    return aesgcm.decrypt(nonce, ciphertext, None).decode()

# Load dataset
dataset = load_dataset("QomSSLab/ara_v7")

# Decrypt rows
dataset = dataset.map(lambda x: {key: decrypt(x[key]) for key in ['AnonymizedJudge_text','AnonymizedJSS_text', 'UniqueTables', 'UniqueItemArray','JSSType','othersdoc']})