Abstract
We describe our deep learning model submitted to the HASOC 2020 shared task on detection of offensive language in social media in three Indo-European languages: English, German, and Hindi. We fine-tune a pre-trained multilingual encoder on the combination of data provided for the competition. Our submission received a competitive macro- average F1 score of 0.4980 on the English Subtask A as well as comparatively strong performance on the German data.
| Original language | English |
|---|---|
| Pages (from-to) | 331-335 |
| Number of pages | 5 |
| Journal | CEUR Workshop Proceedings |
| Volume | 2826 |
| State | Published - 2020 |
| Event | Working Notes of FIRE - 12th Forum for Information Retrieval Evaluation, FIRE-WN 2020 - Hyderabad, India Duration: 16 Dec 2020 → 20 Dec 2020 |
Keywords
- Deep learning
- Multilingual
- Offensive content identification
- Text classification