ARTICLE: Tobias HODEL, "Konsequenzen der Handschriftenerkennung und des maschinellen Lernens für die Geschichtswissenschaft. Anwendung, Einordnung und Methodenkritik" (Historische Zeitschrift CCCXVI (2023) nr. 1, 151-180

 

(image source: DeGruyter)

Abstract:

The ability to read historical manuscripts has been part of the auxiliary scientific method apparatus for centuries; automated handwritten text recognition thus corresponds to a potential facilitation of work that opens up new perspectives for history. In order to assess the technology and its potential, we will explain which factors are crucial for successful recognition and how the results can be further processed. By means of machine or deep learning, however, the method apparatus is also extended by a learning procedure that relies on large amounts of data and adopts valuations from the underlying data, which often leads to undesired side effects. Handwritten text recognition can thus be used to experience and critically evaluate a technology that is currently finding its way into various areas of science and our daily lives.

Read the article here (DOI 10.1515/hzhz-2023-0006).