EGRAPHSEN. Possibilities and perspectives of the digital Painter Attribution for Attic Vases
Cooperation project with the Information Systems and Machine Learning Lab (ISMLL) of the University of Hildesheim (Prof. Dr. Lars Schmidt-Thieme)
The Lower Saxony Ministry of Science and Culture is funding the research project as part of the program „Geistes- und Kulturwissenschaften – digital“.
Contact person: Prof. Dr. Martin Langner
The question of the producers of painted Greek vases has been adressed by archaeology for over 150 years. The computer-assisted analysis of Attic vases now offers the opportunity to put controversial working methods of Classical Archaeology, which were previously based on the expert eye of connoisseurs, to the test. For not isolated criteria, but characteristic combinations of details objectify the allocation of painters, which can be compiled and weighted much more easily by computational methods. In addition, the significance of the archaeological method is debated controversially because the connoisseurship of individual researchers can hardly be objectified as a hermeneutic basis. Therefore, in this project, the assignment of vases to painters, workshops and groups is no longer exemplary and intuitive, but on a broad basis based on data. The aim is to investigate how the attribution criteria are systematised, how they evaluate their relevance, and how the importance of similarity networks is historically weighted by precisely naming criteria and arguments.
For this purpose we would like to develop a data-driven stylometry for Attic vases based on multimodal representations as images and 2D ceramic profiles. In a first step, a selection of vase paintings are annotated by human experts to mark figures, objects and ornaments. Based on these annotations, a deep convolutional neural network is trained. In a second step, deep convolutional neural network architectures are explored, which can predict whether two vase images of human experts have been assigned to the same painter. From the point of view of machine learning, a supervised clustering / record linkage problem for images has to be solved. We will specifically examine models that factor to a high degree through semantic representations, such as scene descriptions and object-class- specific prototypes, that allow us to provide clear explanations for the model's decisions.
For the question of the relevance of the similarity networks for concrete relationships between the vases, the use of computers can also provide impulses because it is possible to describe the type and degree of similarity in a comprehensible way. A further goal is to test digital image classification methods by fundamentally investigating the relationship between archaeological hermeneutics, intuitive connoisseurship and data-based objectification of cognition in a central and intensively researched area of Classical Archaeology.