The Wien[n]erisches Diarium: A digital data trove for the humanities

In the 18th century, the Wien[n]erisches Diarium, still in print today as Wiener Zeitung, was the most important medium of the monarchy. Its complete collection, from 1703 onwards, makes it a source of great scholarly interest – for historical linguistics, media studies, art history, music history, social history and many more.

One goal of this interdisciplinary project (go!digital 2.0) was the digitisation of several hundred 18th century editions in the best possible full-text quality. Where conventional methods of text recognition have so far failed for the German blackletter typeface Fraktur, the tool “Transkribus” – used for the first time outside its originally intended purpose of handwriting recognition – achieved surprisingly good results. By manually correcting more than 6,000 newspaper pages, a model for the Diarium was trained, allowing the self-learning software to automatically recognise further editions with ever increasing accuracy. With good image originals, text accuracy of 99.7 percent can be achieved with this model. This shows how future editions of the Diarium and other historical newspapers can be automatically indexed as full-text.

Over 300 full-text editions of the Wien[n]erisches Diarium are now available online as a data trove in the specially designed web application DIGITARIUM ( digitarium.acdh.oeaw.ac.at , Resch & Kampkaspar 2020) and offer a reliable basis for a wide range of research interests.