Azzindani/ID_REG_KG_2510

Indonesian Legal Regulations - Knowledge Graph Enhanced Dataset

Dataset Description

This dataset contains Indonesian legal regulations enhanced with advanced knowledge graph features for improved Retrieval-Augmented Generation (RAG) systems.

Features

Original Features (from embedding dataset)

• global_id: Unique document identifier
• local_id: Local chunk identifier
• regulation_type: Type of regulation (UU, PP, Perpres, etc.)
• enacting_body: Government body that enacted the regulation
• regulation_number: Official regulation number
• year: Year of enactment
• about: Subject matter of the regulation
• effective_date: Date when regulation became effective
• chapter: Chapter/section information
• article: Article number
• content: Full text content
• chunk_id: Chunk identifier for long documents
• embedding: Semantic embedding vector
• tfidf_vector: TF-IDF vector representation (20,000 dimensions)

Knowledge Graph Features

Entity & Structure Analysis

• kg_entities_json: Extracted legal entities (regulations, organizations, positions)
• kg_entity_count: Number of entities
• kg_cross_references_json: Cross-references to other regulations
• kg_cross_ref_count: Number of cross-references
• kg_legal_domains_json: Legal domain classifications with confidence
• kg_primary_domain: Primary legal domain
• kg_domain_confidence: Confidence score for domain classification

Concept Analysis

• kg_concept_clusters_json: Semantic concept groupings
• kg_cluster_count: Number of concept clusters
• kg_cluster_diversity: Diversity score of concepts
• kg_concept_vector_json: Concept vector for similarity computations

Authority & Hierarchy

• kg_authority_score: Regulatory authority score (0-1)
• kg_hierarchy_level: Position in legal hierarchy (1-10)

Temporal Analysis

• kg_temporal_score: Temporal relevance score with decay
• kg_years_old: Age of regulation in years

Legal Content Analysis

• kg_legal_richness: Density of legal terminology (0-1)
• kg_legal_complexity: Complexity score based on structure (0-1)
• kg_legal_actions_json: Extracted legal action types
• kg_action_types_json: List of action categories
• kg_has_obligations: Boolean - contains obligations
• kg_has_prohibitions: Boolean - contains prohibitions
• kg_has_permissions: Boolean - contains permissions
• kg_sanctions_json: Extracted sanction information
• kg_sanction_types_json: Types of sanctions present

Graph Analytics (if enabled)

• kg_pagerank: PageRank score in citation network
• kg_degree_centrality: Network degree centrality
• kg_citation_impact_json: Citation impact metrics

Quality Metrics

• kg_completeness_score: Information completeness (0-1)
• kg_connectivity_score: Graph connectivity score (0-1)

Use Cases

1. Enhanced RAG Systems: Use graph features for better retrieval ranking
2. Legal Research: Navigate regulation networks via citations
3. Compliance Analysis: Identify related regulations and requirements
4. Temporal Analysis: Track regulation evolution over time
5. Domain Classification: Filter by legal domains
6. Citation Networks: Understand regulatory dependencies

Processing Details

• Source Dataset: Azzindani/ID_REG_Embed_2510
• Processing Method: Chunked streaming (100K rows per chunk)
• Memory Optimization: Disk-based intermediate storage
• Graph Analytics: Citation network with centrality measures
• Entity Extraction: Regex-based NER for Indonesian legal text
• Semantic Analysis: Concept clustering and similarity
• Resume Capability: Checkpoint-based processing

Citation

If you use this dataset, please cite:

@dataset{indonesian_legal_kg_2024,
  author = {Azzindani},
  title = {Indonesian Legal Regulations - Knowledge Graph Enhanced},
  year = {2024},
  publisher = {HuggingFace},
  url = {https://huggingface.co/datasets/Azzindani/ID_REG_KG_Advanced_2510}
}

License

CC-BY-4.0

Contact

For questions or issues, please open an issue in the dataset repository.