--- license: cc-by-4.0 task_categories: - text-generation - summarization language: - id tags: - legal - indonesia - supreme-court - mahkamah-agung - court-decisions - rag --- # βš–οΈ Indonesian Supreme Court HTML Dataset (ID_Supreme_Court_HTML) This repository contains a massive collection of official court decisions from the **Supreme Court of the Republic of Indonesia (Mahkamah Agung RI)**, preserved in their original HTML structure. πŸ›οΈ Unlike plain text datasets, this collection retains the underlying web formatting from the official **Direktori Putusan**, making it an invaluable resource for developers building legal tech tools that require structural context, such as headers, metadata tables, and formal legal hierarchies. πŸ“‚ ## πŸ’‘ The Concept: Structural Legal Preservation Legal documents are more than just text; their meaning is often tied to their layoutβ€”where the **"Menimbang"** (Considering) section ends and the **"MENGADILI"** (Adjudicating) section begins. By providing the data in HTML: * πŸ—οΈ **Structural Parsing:** Developers can use CSS selectors or XPath to precisely extract Case Numbers, Judge Names, and Verdicts. * πŸ” **Rich Context:** Preserves the original tables and lists that are often mangled in PDF-to-text conversions. * πŸ€– **RAG Optimization:** Ideal for Chunking strategies that respect the natural boundaries of a court decision. ## πŸ› οΈ Data Acquisition & Integrity This dataset was compiled using legitimate developer-sanctioned methods: * **Source:** [Mahkamah Agung RI - Direktori Putusan](https://putusan3.mahkamahagung.go.id/) 🌐 * **Methodology:** Systematic retrieval via official web endpoints, adhering to public access guidelines. No aggressive bot-scraping was used; the process was orchestrated to respect the source's stability. βœ… * **Format:** Raw HTML strings mapped to unique decision IDs. ## πŸ§ͺ Use Cases * **Legal LLM Training:** Fine-tune models to understand the specific formal language and structure of Indonesian jurisprudence. πŸŽ“ * **Information Extraction (NER):** Build pipelines to automatically identify the Prosecutor, Defendant, and specific Articles of the law cited. πŸ”Ž * **Semantic Search:** Powering legal search engines that allow users to filter by specific sections of a decision. πŸš€ * **Legal Analytics:** Perform trend analysis on sentencing and judicial reasoning across different regions and years. πŸ“ˆ ## πŸ“‚ Dataset Structure Each entry in the dataset includes: * **ID:** The unique identifier from the Supreme Court directory. * **HTML Content:** The full, raw HTML body of the decision. πŸ“‘ * **Metadata (where available):** Case type (Perdata, Pidana, etc.), year, and court level. ## πŸ“œ License & Attribution This dataset is licensed under the **Creative Commons Attribution 4.0 International (CC BY 4.0)**. πŸ“ ### Attribution 1. **Original Source:** Mahkamah Agung Republik Indonesia. 2. **Dataset Curator:** Azzindani (via Hugging Face Datasets). --- **Disclaimer:** *This dataset is provided for research and development purposes. It is not an official legal record. For official legal proceedings, always refer to the verified physical or digital documents provided by the Indonesian Supreme Court.* ⚠️ ---