In Ukraine, a collection of handwritten texts has been initiated to expand the specialized RUKOPYS dataset. The initiative aims to collect as many samples of Ukrainian writing as possible in order to train artificial intelligence models to qualitatively recognize both modern and archival handwriting.
In Ukraine, a collection of handwritten texts has been initiated to expand the specialized RUKOPYS dataset. The initiative aims to collect as many samples of Ukrainian writing as possible in order to train artificial intelligence models to qualitatively recognize both modern and archival handwriting.
The Ministry of Economy of Ukraine announced the start of the materials collection campaign.
Libraries, museums, public and private archives, as well as educational institutions — from schools to universities — that can provide exam papers, homework assignments, or student notes for digitization are invited to cooperate. In addition, public organizations and ordinary Ukrainians who keep family archives, old letters, diaries, or notes at home can also join.
The developers emphasize that the effectiveness of the models directly depends on the diversity of the data: the more eras, sources, and unique writing styles that enter the database, the stronger the ecosystem of solutions for automatic recognition of Ukrainian texts will become. The organizers fully take on the technical process of digitizing and processing the provided materials.
Legal support for the initiative is provided by the Ukrainian Catholic University (UCU), which acts as the publisher of the database. After formation, the dataset will be published in the open access under the CC BY-NC-SA 4.0 license with mandatory indication of all sources and authors.
Earlier, Dmytro Voytech, Advisor on Artificial Intelligence at the Ministry of Economy, said in the AI&I podcast that the initiative will significantly accelerate the digitalization of public services and open the way to the global digitization of historical archives.
