The goal in Kallimachos is to build a complete text analysis pipeline, starting with OCR from paper and going up to high-level text mining. We are mostly concerned with the later steps in this pipeline, performing various machine learning tasks on historical novels as well as modern texts. To this end, we employ state-of-the-art techniques like deep neural networks and develop new models that help us better understand the narrative of novels.
So far, we have published papers on the detection of literary subgenres from novels and built towards a computational representation of literary plot by automatically identifying one important plot element, namely happy endings. We have also found that emotions play an important role in characterising the plot of a novel and are therefore working on bringing Sentiment Analysis to the domain of German literature. Sentiment Analysis can be used to build "trajectories" over the story of a novel, which can then be used to identify important events by looking at emotional peaks (see this repository).
Recently, we have tried to automatically identify direct speech in novels, which is useful to characterise the relationship between different characters in a novel.
If this sounds interesting to you, do not hesitate to contact us, we are always offering Bachelor/Master Theses and practicals.
The following persons are or were involved in this project:
Here is a list of recent publications from the project. For a full list, please see here.
Gius, E., Jannidis, F., Krug, M., Zehe, A., Hotho, A., Puppe, F., Krebs, J., Reiter, N., Wiedmer, N., & Konle, L. Detection of Scenes in Fiction. Proceedings of Digital Humanities 2019.
Jannidis, F., Konle, L., Zehe, A., Hotho, A., & Krug, M. Analysing Direct Speech in German Novels. DHd 2018.
Zehe, A., Becker, M., Jannidis, F., & Hotho, A. Towards Sentiment Analysis on German Literature.