“The Rixdorf Sermons” - manuscript identification and linguistic author detection
At the Institute of Slavic Studies of the Humboldt University of Berlin, 18th century manuscripts in the Czech language from Böhmisch-Rixdorf in Berlin are closely studied and computer-aided analysis programs are made use of to find out how language change and cultural exchange can be demonstrated. This interdisciplinary research project is funded by the Volkswagen Foundation as part of the "Mixed Methods" initiative in the humanities.
The main focus lies on 5,000 manuscript pages written between 1740 and 1830. Those documents record the life of a small, originally Czech-speaking community of religious refugees from Bohemia. The community was founded by about 350 Hussite believers who fled from East Bohemia to their new home in Berlin-Rixdorf in 1737. Their records are kept in an archive in Berlin-Neukölln.
The interaction of both methodologies – text analysis and image / pattern recognition – shall make it possible to find and test significant features for the following tasks:
The focus of the task of the MusterFabrik Berlin here is to develop methods for the analysis of the context-specific structure of image patterns in the digitized manuscripts in order to derive qualitative statements about the authors and the contents of the sermons.
For this purpose, the development and application of artificial intelligence shall detect recurring image patterns in the digitized data – such as contextual connections in form of letters, words or phrases – and structurally record their positional occurrence. A so-called structured network of image patterns shall be derived, which describes the detected recurring image patterns, their number and their positional connection with each other. As a result, a qualitative analysis of the manuscripts shall be carried out in cooperation with experts from the Humboldt University of Berlin. The context-specific relationships already revealed with the aid of machine learning methods will be analyzed with the aim to determine whether qualitative statements can be derived from them.
The first step towards such a structural analysis is to detect image features suitable for structural evaluation. Based on this, a module for the detection of image patterns is to be developed. Furthermore, machine learning methods for a context-specific analysis and for making positional connections shall also be developed. Finally, a qualitative analysis will be carried out in cooperation with staff of the Humboldt University of Berlin with regard to the possibility of deriving qualitative statements from the results obtained.
The project is funded by the Volkswagen Foundation: