Research interests
Hate Speech and Offensive Language in Social Media
Recently I have been investigating identitiy term bias in hate speech detection and develop debiasing methods.
Co-organization of the FIRE 2021 shared task HASOC subtrack: Hate Speech and Offensive Content Identification in English and Indo-Aryan Languages. [Website]
Co-organization of the FIRE 2020 shared task HASOC subtrack: Hate Speech and Offensive Content Identification in Indo-European Languages. [Website]
Experiments with hate speech detection in Tweets and identification of potentially illegal hate speech in the context of the projects DTCT and HASeKI (referenced below).
Research on contexts of offensive language in conversations on reddit. [Offensive Language in Dialogues]
Integrating Explicit Knowledge into Neural Network Approaches
The main topic of my current research. The underlying motivation is to integrate acquired knowledge about a specific task into state-of-the-art neural classification approaches. I mainly experimented on the task of hate speech detection, some of my publications point out early results.
Terminology Extraction
In the project "Textanalyse und Ontologieaufbau" (2014-2017) at the University of Stuttgart in cooperation with the Robert Bosch GmbH we investigated statistical methods to automatically extract domain-specific terminological expressions from text corpora. In the context of this research I wrote my Bachelor's and Master's thesis.
In cooperation with Tim van Niekerk from the Dictionary Unit for South African English (DSAE) we developed at the University of Hildesheim methods to extract uses of language-specific terms from South African English corpora (2017-2018). [Website]
Other Text Analyses
Various analyses of (German/English) text corpora, social media data, text classification, supervised machine learning, multi-modal data, ...
Projects
Experiments on hate speech detection in the project "Das Phänomen Hate Speech und seine Erkennung durch KI: interdisziplinär – international – erklärbar? (HASeKI)" (2020-2021). [Website]
Co-developing the hate speech detection system in the project "Detect Then Act (DTCT) - Taking Direct Action against Online Hate Speech by Turning Bystanders into Upstanders" (2019-2021). [Website]