Принстонский проект «Джениза» опубликовал полный набор данных для исследований в области искусственного интеллекта

Journal of Cultural Analytics publishes data essay on Cairo Geniza, bridging medieval Middle East with data science and AI 1 September 2026 The Center for Digital Humanities (CDH) at Princeton University announces the publication of "Princeton Geniza Project Datasets," a new data essay by Rebecca Sutton Koeser (CDH) and Marina Rustow (Near Eastern Studies/History), published in the Journal of Cultural Analytics (Koeser and Rustow 2026, 11(3): 1–35, DOI: 10.22148/jca.1084). The essay presents, for the first time, a full account of the data behind the Princeton Geniza Project (PGP) — one of the richest digital resources for the study of the medieval Islamic and Jewish Mediterranean world. The PGP dataset opens rich and complex historical material to data science approaches The Cairo Geniza — a repository of discarded largely Hebrew-script writings in the Ben Ezra Synagogue in Fustat, Old Cairo — preserved hundreds of thousands of documents from everyday medieval life c. 950–1250, from personal letters and legal contracts to trade accounts. These fragments offer scholars an unusually candid window onto medieval culture stretching from Spain to Sumatra. Since the 1980s, Princeton’s Geniza Project has been collecting, documenting and annotating these fragments, which are now scattered across some seventy libraries worldwide. Today, the PGP contains 35,855 Geniza documents — with descriptions, and information about document types, languages and scripts, dates across multiple calendars, transcriptions, translations, and bibliographic citations — along with information on 1,802 people and 486 places. The publication of the dataset – a complement to the public-facing PGP website – makes the complex material accessible for data science and AI researchers and historians alike. Project PI Marina Rustow’s collaboration with the Center for Digital Humanities has made the PGP one of the leading efforts bringing medieval manuscripts into the world of data science and AI — alongside projects such as the multi-institutional OpenITI project for Arabic texts and the National Library of Israel's KTIV for Hebrew manuscripts. The essay documents the PGP’s 40-year history — from an IBM-funded 1986 pilot to the five-and-a-half-year PGP–CDH partnership (2020–2025) that rebuilt the database from the ground up — and explains the modeling decisions behind the data, including the many-to-many relationship between physical fragments and the documents written on them. "Providing both a well-structured dataset, as well as the contextual information of its creation, enables a broader community of researchers to engage with the Geniza material," states the authors. "The study of the medieval Mediterranean world can now be explored, interrogated, and modeled — inviting the tools of data science and AI to reveal patterns across hundreds of thousands of fragments that no single scholar could trace by hand." About the Dataset The Princeton Geniza Project dataset is archived on Zenodo (DOI: 10.5281/zenodo.18716581) and updated quarterly. The essay itself is open access under a Creative Commons Attribution 4.0 license. The Princeton Geniza Project is the flagship project of the Princeton Geniza Lab (PGL), directed by Marina Rustow; PGL is an affiliate lab of the Center for Digital Humanities. Read the essay: https://doi.org/10.22148/jca.1084