Digitisation
(image generated with Artificial Intelligence)
We digitise a set of over a hundred thousand folios using the unique expertise of a dedicated external partner. This will lead to an open access corpus publication of the tagged manuscript, using Named Entity Recognition. Countless doctrinal treatises, pamphlets, newspapers have been digitised in the past quarter century. With the development of handwritten text recognition (HTR), the 'mountain of paper' (or the kilometres of letters and memoranda conserved in Europe's archives) can be added as an essential complement to sources circulating in the early modern public sphere. The history of international can move beyond 'big names' and treaties, but access to the colossal corpus of precious handwritten material is often constrained by opening hours or paleographic difficulties.
The use of algorithms can speed up HTR (= the identification of graphical elements as letters or numbers, akin to 'OCR' for print material), but cannot be scientifically credible without human judgment. 'AI' always makes mistakes. Hence, the rule of the team members in correcting automated transcriptions, tagging, disambiguating and making links to external repositories and databases (VIAF, Wikidata, Prosocour, PaPa, Europeana [Gallica, Bayerische Staatsbibliothek, Österreichische Nationalbibliothek, KBR], Google Books, Goetgevonden...) is crucial. A purely quantitative approach to the material would be highly unsuitable. Precisely due to the inevitable presence of transcription errors, the digitisation will provide the reader with the chain of evidence: original image and machine-readable text will be side by side.
During the project's lifespan, we shall report on intermediary findings, notably with regards to the methodological caveats of classical 'historical criticism' applied to the digital environment. Likewise, the tagging of the manuscript through categories of persons, places, dates, treaties, institutional concepts and organisations.