Semantic Drift in Long-Form Financial Disclosures in Portuguese: Disclosure PDF Corpus

Corpus of public Portuguese financial disclosure PDFs used in the article Semantic Drift in Long-Form Financial Disclosures in Portuguese, published online as a journal pre-proof in The Journal of Finance and Data Science on July 2, 2026 (article 100199, DOI: 10.1016/j.jfds.2026.100199). The archive contains five per-dataset zip files: Vale, Itau, EDP, Conab graos, and SLC Agricola. The repository provides the analysis code, manifest, checksums, and reference result tables. The PDF files are public regulatory filings. Source terms govern the raw documents. The accompanying manifest and checksums support reproducibility.