# Corpus configuration — WP45 / JoF-250
corpus_root: "~/Desktop/Journal-of-Finance-2018-2020"   # original PDFs, read-only, never modified
bibliographic_index: "~/Desktop/Journal-of-Finance-2018-2020/index.csv"  # produced by the Safari/UQO-proxy download (2026-11-03)
journal: "The Journal of Finance"
issn: "1540-6261"
years_in_scope: [2017, 2018, 2019, 2020]   # all pre-ChatGPT (public release 2022-11-30); 2017 present in folder and included
article_sections_in_scope: ["ARTICLES", "REPLICATIONS AND CORRIGENDA", "CORRIGENDUM", "THIS ARTICLE HAS BEEN RETRACTED"]
derived_dir: "data/derived"
manifests_dir: "data/manifests"
min_pdf_bytes: 20000
crossref_verify: true          # free public API, no key; polite pool with mailto in config/models.yaml