B6, 26, Room C 1.05
Focus group: Natural language processing
I am a Ph.D. candidate under the supervision of Prof. Simone Paolo Ponzetto and Prof. Goran Glavaš working on neural methods for natural language understanding. My research centers around two main topics: Argument Mining and Language Representations. In Argument Mining, the overall goal is to teach machines to understand human argumentation. This involves several challenging tasks such as identifying argumentative units (e.g., claims, premises) and relationships (e.g., attack, support) and predicting the argumentative quality of texts. Given that we can understand scientific publications as arguments, I also work on solving various problems in the area of scientific publication mining (with Prof. Eckert (Stuttgart Media University)). In this context, I am trying to obtain a deeper insight into the argumentative nature of scientific writing by developing state-of-the-art methods for analyzing and exploring the rhetorical aspects of scientific publications, which we collectively dub 'scitorics'.
The key to being successful in computationally analyzing argumentation is employing high-quality language representations. To this end, I also work on the specialization of language representations, for instance, by injecting linguistic or common sense knowledge into text encoder models. In this context, I am also trying to understand how ethically fair text representations should look like, for example, by detecting unfair biases such as sexism and racism in our models and by developing debiasing algorithms.
Formerly, I was funded by the project „Linked Open Citation Database (LOC-DB). You can find more information about the project here: https://locdb.bib.uni-mannheim.de.