Mirna2Nageh/Egyptian_Criminal_Legal_Assistant_RAG_V2

Egyptian Criminal Legal Assistant Dataset (ECLA-DB)

Dataset Summary

ECLA-DB is a comprehensive Arabic legal corpus specifically curated for building Retrieval-Augmented Generation (RAG) systems targeting Egyptian Criminal Law. The dataset aggregates 1,058 documents across 91 legally-categorized directories, covering the full spectrum of Egyptian criminal jurisprudence — from primary legislation and Court of Cassation rulings to forensic science literature and procedural defense encyclopedias.

This dataset was constructed to power an AI Legal Assistant capable of:

1. Answering legal queries grounded in Egyptian criminal statutes and precedents
2. Analyzing case files to detect procedural violations and legal loopholes
3. Generating defense memoranda in formal legal Arabic (فصحى)

---

Dataset Statistics

Metric	Value
Total Files	1,058
Total Directories	91
Primary Language	Modern Standard Arabic (Legal Register)
File Formats	.txt, .pdf, .docx, .doc
Legal Domain	Egyptian Criminal Law
Geographic Scope	Arab Republic of Egypt
Temporal Coverage	Court rulings spanning multiple decades

---

Dataset Structure

The corpus is organized into five primary legal pillars, each subdivided into granular thematic directories:

1. Criminal Procedure Law (اجراءات جنائية)

Covers the full lifecycle of Egyptian criminal proceedings.

Sub-collection	Description
Core Procedure Codes	Primary legislation texts (قانون الإجراءات الجنائية) — multiple annotated editions
Al-Marsafawi Encyclopedia	موسوعة المرصفاوي في أصول الإجراءات الجنائية (Vols. 1–2)
Gendy Abdel Malek Encyclopedia	موسوعة المستشار جندي عبد الملك (Vols. 1–5)
Procedural Commentary	Scholarly commentaries by leading Egyptian legal academics

2. Court of Cassation Rulings (محكمة النقض + القواعد التي قررتها محكمة النقض)

The most legally significant component of the dataset.

Sub-collection	Description
محكمة النقض/	78 individual Cassation rulings (.txt and .pdf)
القواعد التي قررتها محكمة النقض	Compiled legal principles — 4 volumes
المستحدث من المبادئ	Recent doctrine from Criminal Cassation Chambers
موسوعة أحكام النقض الجنائية	Thematically indexed Cassation rulings — largest sub-collection
مجموعة عمر الجنائية	Omar Criminal Collection (Vols. 1–7)
مجموعة القواعد القانونية	Legal rules collection (Vols. 1–2 at root level)

3. Criminal Court Cases (جنايات)

51 real Egyptian criminal court case files.

Property	Detail
Case count	51 files (case1 → case51)
Format	Mixed .txt and .pdf
Source	Egyptian Criminal Courts (Janayet)
Use in RAG	Pattern matching, defense style templates, Few-Shot examples
RAG Label	past_case (NOT treated as legal authority)

4. Forensic Medicine (سلسلة الطب الشرعي)

A specialized sub-corpus for forensic science queries.

File	Topic
إصابات الأسلحة النارية	Firearm injuries
الاخطاء الطبية	Medical malpractice
الاغتصاب الجنسى	Sexual assault forensics
التفجيرات الارهابية	Terrorism forensics
جرائم التعذيب	Torture crime evidence
حوادث الطرق	Traffic accident forensics
معاينة مسرح الجريمة	Crime scene examination

5. Criminal Law Reference (قانون الجنائي)

Scholarly works and specialized studies.

Category	Examples
Defense Pleas	أهم الدفوع الجنائية، دفوع قضايا المخدرات
Criminal Intent	القصد الجنائي، سبق الإصرار
Specific Crimes	جرائم الضرب والجرح، جرائم الكمبيوتر
Procedural Rights	حقوق المتهم، الإخلال بحق الدفاع

6. موسوعة أحكام النقض الجنائية — Thematic Cassation Index

This is the most granular sub-collection: Cassation rulings indexed by crime type and legal concept across 50+ thematic folders.

Sample thematic folders: تفتيش · قتل عمد · قتل خطأ · دفوع · دفاع · إثبات · تزوير · رشوة · مخدرات · إجراءات المحاكمة · حكم · تحقيق · قبض · بطلان · اختصاص ...and 35+ more

---

Intended Use

Primary Use Case

Building RAG pipelines for Egyptian criminal legal AI assistants.

Recommended RAG Index Architecture

Vector Store Collections:
├── legislative      ← Primary law texts + Constitution
├── naqd_rulings     ← All Cassation rulings & principles
├── reference_books  ← Encyclopedias + Forensic medicine
└── past_cases       ← 51 criminal court cases (pattern matching only)

Recommended Metadata Schema per Chunk

{
  "source_type": "naqd_ruling | legislative | reference_book | past_case",
  "doc_title": "string",
  "legal_category": "criminal | procedural | forensic | constitutional",
  "crime_type": "string | null",
  "year": "integer | null",
  "court_name": "string | null",
  "section": "string",
  "page_number": "integer",
  "chunk_id": "string"
}

---

Data Collection & Processing

Sources

• Digitized Egyptian legal archives
• Rare legal encyclopedias (Gendy Abdel Malek, Al-Marsafawi, Omar Criminal Collection)
• Egyptian Court of Cassation published rulings
• University legal libraries (Faculty of Law, Cairo University)

Processing Pipeline

1. OCR: Scanned PDFs were processed via Google Docs OCR engine (superior Arabic contextual correction vs. Tesseract)
2. Format Normalization: All documents converted to UTF-8 encoded plain text
3. Structural Organization: Manual taxonomic classification into 91 legal directories
4. Quality: Text-readable (no binary OCR artifacts in .txt files); PDFs may contain minor encoding variance

Known Limitations

• Some PDF files may contain residual OCR artifacts from the conversion pipeline
• Temporal coverage of Cassation rulings is not uniformly dated across all files
• The past_cases collection should NOT be used as legal authority — only as stylistic/structural reference

---

Citation

If you use this dataset, please cite:

@dataset{ecla_db_2025,
  title        = {Egyptian Criminal Legal Assistant Dataset (ECLA-DB)},
  author       = {Mirna Nageh Botros},
  year         = {2025},
  note         = {Graduation project dataset — Egyptian Criminal Law RAG System},
  language     = {Arabic},
  howpublished = {\url{https://huggingface.co/datasets/Mirna2Nageh/ECLA-DB}}
}

---

License

This dataset is released under CC BY-NC 4.0 (Creative Commons Attribution-NonCommercial 4.0).
It may be used for research and educational purposes. Commercial use is prohibited without explicit permission.

---

Contact

For questions regarding the dataset or the AI Legal Assistant project, please open an issue on the associated GitHub repository.