Tianyige Ming Examination Records: Full Text and English Translation

Machine transcription, punctuation and English translation of 312 printed records of the Ming dynasty civil-service examinations (科舉錄), digitized by the Tianyi Pavilion Museum (天一閣博物院, Ningbo) and published on Wikimedia Commons as public domain. The set covers 236 provincial (鄉試錄), 37 metropolitan (會試錄) and 39 palace (進士登科錄) examination records from 1371 to 1684: 40,461 page images and about 9.9 million characters. For every page the dataset gives the transcription, a link to the scan, modern punctuation and an English translation. Where two models read a page (about 12,600 pages), it also gives the second reading and an agreement score. Each record is also split into its logical documents: - prefaces and lists of officials - 12,921 examination questions - 49,378 graduate entries - 7,348 model answers (程文), with the examiners' comments (批語) and a link to the question each answers Transcription was done by vision-language models (gemini-4-argon and Claude Opus 5.5), checked against volunteer transcriptions from Chinese Wikisource at about 2.3–2.4% character error rate. Most of that residual is variant character forms or reference errors; the estimated true error on essay text is about 1%. Translation and segmentation were done by gpt-6-astra and gemini-4-argon. All text is uncorrected machine output and should be checked against the linked scans for citation. Contents: per-page and per-document JSON Lines files, tables of questions and graduates (TSV and JSON Lines), record metadata, plain-text files per record, and a static browser (index.html). See README.md, in English and Chinese, for full documentation.