AI Wizards at CheckThat! 2025: Enhancing Transformer-Based Embeddings with Sentiment for Subjectivity Detection in News Articles

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Fasulo, Matteo, Babboni, Luca, Tedeschini, Luca
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916962914795520
author Fasulo, Matteo
Babboni, Luca
Tedeschini, Luca
author_facet Fasulo, Matteo
Babboni, Luca
Tedeschini, Luca
contents This 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 Arabic, German, English, Italian, and Bulgarian; final evaluation included additional unseen languages (e.g., Greek, Romanian, Polish, Ukrainian) to assess generalization. Our primary strategy enhanced transformer-based classifiers by integrating sentiment scores, derived from an auxiliary model, with sentence representations, aiming to improve upon standard fine-tuning. We explored this sentiment-augmented architecture with mDeBERTaV3-base, ModernBERT-base (English), and Llama3.2-1B. To address class imbalance, prevalent across languages, we employed decision threshold calibration optimized on the development set. Our experiments show sentiment feature integration significantly boosts performance, especially subjective F1 score. This framework led to high rankings, notably 1st for Greek (Macro F1 = 0.51).
format Preprint
id arxiv_https___arxiv_org_abs_2507_11764
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AI Wizards at CheckThat! 2025: Enhancing Transformer-Based Embeddings with Sentiment for Subjectivity Detection in News Articles
Fasulo, Matteo
Babboni, Luca
Tedeschini, Luca
Computation and Language
Information Retrieval
This 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 Arabic, German, English, Italian, and Bulgarian; final evaluation included additional unseen languages (e.g., Greek, Romanian, Polish, Ukrainian) to assess generalization. Our primary strategy enhanced transformer-based classifiers by integrating sentiment scores, derived from an auxiliary model, with sentence representations, aiming to improve upon standard fine-tuning. We explored this sentiment-augmented architecture with mDeBERTaV3-base, ModernBERT-base (English), and Llama3.2-1B. To address class imbalance, prevalent across languages, we employed decision threshold calibration optimized on the development set. Our experiments show sentiment feature integration significantly boosts performance, especially subjective F1 score. This framework led to high rankings, notably 1st for Greek (Macro F1 = 0.51).
title AI Wizards at CheckThat! 2025: Enhancing Transformer-Based Embeddings with Sentiment for Subjectivity Detection in News Articles
topic Computation and Language
Information Retrieval
url https://arxiv.org/abs/2507.11764