Smit, Jinna: Automatic writer identification: the paleographer’s new best friend?

Posted by GV in Paper, Proposals |

Paleographical analysis is a tried and tested method used for a number of research questions. Its application is limited though for it is not only very time-consuming but also subjective at heart. Luckily, these aspects have not prevented paleographers to carry on with their studies, but with new technologies available,  one cannot help looking for a more efficient and objective method. The first part of this paper deals with such a quest, which started with 600 early fourteenth-century charters of the chancery of the counts of Holland/Hainault and resulted in the application of the Groningen Intelligent Writer Identification System (GIWIS). Initially, GIWIS was developed by the Rijksuniversiteit Groningen to facilitate the identification of modern writing hands, but after some tweaking, the system can now process medieval charters as well. Far from being the first automatic writer identification system, its unique features even make GIWIS highly suitable to be applied on medieval documents. During this presentation I will describe the techniques employed by GIWIS, the required input of the researcher, my test results and their reliability. Subsequently, I will discuss those cases in which computer and paleographer disagree and the implications for future research.

Jinna Smit, PhD researcher
University of Amsterdam / Nationaal Archief

Comments are closed.