Czech Legal & Legislative Text Dataset for AI/LLM Training (1918–2026)

A structured corpus of Czech legislation, extracted directly from the Czech Republic’s official open data portal for legal texts. Built for AI/LLM training, legal NLP research, and Czech language corpus development. Legal basis: Underlying statutory text is excluded from copyright protection under Act No. 121/2000 Coll. Official government data is published under an open data mandate (Act No. 106/1999 Coll.) with commercial reuse explicitly permitted. Scope: 229,293 individual legal provisions (§, Čl., or equivalent) across 20,600 distinct Czech legal acts, spanning 1918 to present. 27,494,397 words (188,529,608 characters) of source text. Language: 96% Czech, 4% Slovak. The Slovak content is genuine material inherited from the shared Czechoslovak federal era legal collection (1918 to 1938 and 1945 to 1992). Every Slovak tagged row was checked to fall within these periods. Structure: Each row contains the latest consolidated wording of the provision, alongside effective dates, amendment count, repeal status and date, document type (Act, Decree, Regulation, etc., in Czech and English), and normalized collection metadata (Sbírka zákonů / Úřední list). The corpus includes both acts in force and repealed acts; repealed acts carry their final wording and are flagged, so buyers can filter for currently valid law. Coverage note: e-Sbírka’s full catalog lists 46,039 acts. Of these, 15,487 are placeholder entries with no text of their own, and 9,465 (20.6%) contain only amendment inst