A new digital tool developed at Vilnius University in Lithuania is expected to help researchers, archives and individuals unlock manuscript and historically printed documents written in Yiddish and Hebrew.
The service, known as VILNISH, uses artificial intelligence to recognize handwritten and printed historical texts and convert them into searchable digital form. It builds on the Vilne-Yiddish language model developed by researcher Sergii Gurbych, whose work focuses on applying machine learning to Jewish archival sources.
The technology is designed to process manuscripts that have long posed challenges for historians, including handwritten diaries, letters, synagogue records and other community documents. By transforming scanned pages into machine-readable text, the tool allows users to search, index and analyze materials that previously required time-consuming manual transcription.
VILNISH can recognize both handwritten and historical printed texts while preserving their original spelling and orthographic features. In addition to text recognition, the service can adapt and translate the material into English or Lithuanian, making the sources accessible to researchers who do not read Yiddish or Hebrew.
The platform is intended for a wide range of users. Archives, libraries and museums can use it to digitize and catalog large collections of historical manuscripts, enabling full-text searches across their holdings. Scholars working in Jewish history and digital humanities can apply the tool to analyze large corpora of documents more efficiently.
The service may also benefit private individuals researching family history, allowing them to decipher letters or records written generations ago in Yiddish or Hebrew. In some cases, such documents can also serve as supporting evidence in legal matters, including applications for citizenship or repatriation.
Specialists at the university’s Center for the Study of East European Jewish History process materials submitted to the service and adapt the recognition system to the characteristics of each collection, improving accuracy across different handwriting styles and historical sources.
