Retrievability and archival visibility in RTP Arquivos: dataset and scripts

Companion dataset and code for the article "Retrievability and archival visibility in a public broadcaster's online archive: lexical search, controlled indexing and the uneven documentation of the Lusophone space in RTP Arquivos" (submitted to the Journal of Documentation, 2026). RTP Arquivos is the online archive of Rádio e Televisão de Portugal, Portugal's public service broadcaster. The deposit contains: (1) the enumeration of the portal's 187,179 content URLs from its XML sitemaps (27 October 2026); (2) a simple random sample of 3,000 items with metadata, used as a portal-wide baseline and for recall estimation; (3) three query-based collections retrieved through the portal's own search for the country names and demonyms of Angola, Brazil, Mozambique, Cape Verde, Guinea-Bissau, São Tomé and Príncipe, Timor-Leste and Macau (9,668 + 3,562 + 2,850 items); (4) the deduplicated corpus of 14,045 unique items with the analytical groups through which each was retrieved; (5) the collection, analysis and known-item-probe scripts; and (6) a script (reproduce_tables.py) that reproduces the article's tables from the deposited files alone. Each item carries its URL, title, date, programme and series names, the archive's structured fields (Locais, Personalidades, Temas, channel, content type) and derived boolean variables (presence of lexical stems in the item's own record; presence of each place in the Locais field; presence of an analytical summary). The archive's descriptive prose (descriptions, summaries, full page text) is RTP's copyrighted content and is not redistributed; every item remains consultable at its URL. Data and documentation: CC BY 4.0; scripts: MIT.