This innovative tool revolutionizes palaeographical research in Hittite studies, providing quick and convenient access to palaeographic tables based on photographs and, in the near future, also 3D models. The current version offers access to more than 5 million attestations of cuneiform signs.

Development and Functions

The Palaeographicum (accessible through the Konkordanz der hethitischen Keilschrifttafeln) provides support for rapid location of diagnostic cuneiform characters for dating texts or comparing manuscripts by extracting cuneiform characters from photographs of clay tablets and compiling them in tables for comparison. This largely replaces the time-consuming and laborious search for comparative characters.

The first beta version offers 3.5 million excerpts of cuneiform characters on black-and-white photos and 1.5 million colour excerpts.

The basis for this tool is the AI model developed in the CuKa project (Computer-assisted cuneiform analysis 2018 to 2023, in cooperation with Gernot Fink and Frank Weichert, TU Dortmund), which was funded by the German Research Foundation (DFG, project number 405966540). This project developed a demonstrator, a web-based, collaborative visualisation, analysis and research system for evaluating cuneiform tablets using photographs (2D) and 3D scans.

The concept comprises three search methods: (1) Query-by-Example (QbE): Search based on a marked image section; the AI model finds visually similar characters. (2) Query-by-String (QbS): Search using the standard character name in cuneiform studies. (3) Query-by-Expression (QbX): Search using the structural properties of a character (geometric wedge arrangement). Regardless of the usual line-based approach, the search is performed according to region proposals.

The philological work was carried out and supervised by Turna Somel (Academy of Sciences and Literature, Mainz), who annotated the training material and evaluated the training results. Christopher Rest and Eugen Rusakov (both TU Dortmund) were responsible for developing the AI model.

The demonstrator's visual online tool proved too cumbersome and complicated to process the vast amount of available visual material. The prototype was therefore modified and further developed by G.G.W. Müller after completion of the project, resulting in the first version of the Palaeographicum. We would like to thank Christopher Rest and Herbert Baier Saip (ZPD, University of Würzburg) for their technical support.

Financial and Institutional Support