The Consortium Profiling Premodern Authors (PROPREAU) applies and develops new flexible machine learning-based tools for the analysis and classification of texts in order to explore several fundamental and unresolved questions of authorship in classical and medieval Latin texts. Despite the cultural importance of Latin, many essential texts remain anonymous. This is largely so because of the highly conservative nature of Latin based culture, characterized by imitation of earlier authors and quoting excerpts of their texts. Therefore, authorship attribution requires an analysis and comparison of large quantities of text. PROPREAU incorporates machine learning-based tools developed at the Turku NLP Group (IT Department) into the conventional argumentation of the humanities, allowing a much wider look at textual material than is attainable by single scholars using conventional methods.
The expected results of the Consortium are new, well-grounded answers to questions of authorship that were previously considered unsolvable. PROPREAU will provide guidelines and new computational methods for future endeavors to identify anonymous premodern texts. The subproject at the Department of Cultural History addresses several significant aspects of the Latin literary culture: reconsidering and re-attribution of Ciceronian corpus, patristic texts and their medieval imitation, forgery and authencity of early-medieval works, and the identification of late-medieval polemical treatises.
The researchers of the Turku NLP Group are responsible for the computational analysis, using both a simple but effective linear classifier, Support Vector Machine (SVM), as well as the Convolutional Neural Network (CNN), representing the current state of the art in text classification. Both classifiers have been proven effective in the attribution tasks of modern texts, but their implementation to ancient and medieval works requires new solutions.
The Consortium is funded by the Academy of Finland, Digital Humanities Academy Programme (2016-2019)
tel. +358 2 333 5054