Search:
Match:
1 results

Analysis

This paper addresses the under-representation of hope speech in NLP, particularly in low-resource languages like Urdu. It leverages pre-trained transformer models (XLM-RoBERTa, mBERT, EuroBERT, UrduBERT) to create a multilingual framework for hope speech detection. The focus on Urdu and the strong performance on the PolyHope-M 2025 benchmark, along with competitive results in other languages, demonstrates the potential of applying existing multilingual models in resource-constrained environments to foster positive online communication.
Reference

Evaluations on the PolyHope-M 2025 benchmark demonstrate strong performance, achieving F1-scores of 95.2% for Urdu binary classification and 65.2% for Urdu multi-class classification, with similarly competitive results in Spanish, German, and English.