Venue: CEUR Workshop Proceedings (EXIST 2023 shared task), 13 pp.
Authors: J. Böck, M. Schütz, D. Liakhovets, N. Q. Satriani, A. Babic, D. Slijepčević, M. Zeppelzauer, A. Schindler
This paper describes the AIT_FHSTP team's system for the EXIST 2023 benchmark on sexism detection in social media. The approach combines transfer learning with sentiment and toxicity embeddings and hand-crafted features, blending modern transformer-based representations with targeted, interpretable signals to improve detection.
The work fits into my broader research on detecting harmful content online.
BibTeX and details: https://icmt.ustp.at/bibtex/download/69Z4BZYS