Optimism, Expectation, or Sarcasm? Multi-Class Hope Speech Detection in Spanish and English

2025-04-24

AI Summary

Hope is a complex, polysemous, and under-studied affective state; existing work lacks fine-grained, multilingual recognition frameworks distinguishing its semantic subtypes. Method: We introduce a multilingual fine-grained hope identification task for English and Spanish, categorizing utterances into four types: generic hope, realistic hope, unrealistic hope, and ironic hope. We construct PolyHope V2—the first bilingual annotated dataset (30K+ tweets)—and propose a four-dimensional explicit annotation schema, the first to systematically model ironic hope. We comparatively evaluate fine-tuned BERT/RoBERTa against zero-/few-shot prompting of GPT-4/Llama-3, employing confusion matrix analysis and qualitative inspection to uncover systematic confusions among semantically proximal classes. Contribution/Results: Fine-tuned Transformer models substantially outperform LLM prompting, especially in ironic hope detection and subtype discrimination. PolyHope V2 and our benchmark establish a new foundation for multilingual affective computing with heightened semantic sensitivity.

Technology Category

• Natural Language Processing: Fact-Checking / Misinformation Detection (NLP Focus)
• Machine Learning: Large Multimodal Models (LMMs)
• Cognitive Modeling & Cognitive Systems: Affective Computing

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Mining multimedia, multimodal, multilingual, cross-lingual Web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Hope is a complex and underexplored emotional state that plays a significant role in education, mental health, and social interaction. Unlike basic emotions, hope manifests in nuanced forms ranging from grounded optimism to exaggerated wishfulness or sarcasm, making it difficult for Natural Language Processing systems to detect accurately. This study introduces PolyHope V2, a multilingual, fine-grained hope speech dataset comprising over 30,000 annotated tweets in English and Spanish. This resource distinguishes between four hope subtypes Generalized, Realistic, Unrealistic, and Sarcastic and enhances existing datasets by explicitly labeling sarcastic instances. We benchmark multiple pretrained transformer models and compare them with large language models (LLMs) such as GPT 4 and Llama 3 under zero-shot and few-shot regimes. Our findings show that fine-tuned transformers outperform prompt-based LLMs, especially in distinguishing nuanced hope categories and sarcasm. Through qualitative analysis and confusion matrices, we highlight systematic challenges in separating closely related hope subtypes. The dataset and results provide a robust foundation for future emotion recognition tasks that demand greater semantic and contextual sensitivity across languages.

Problem

• Detecting nuanced hope speech types in multilingual texts
• Distinguishing sarcasm from genuine hope in social media
• Improving NLP model accuracy for fine-grained emotion classification

Innovation

• Multilingual dataset with nuanced hope subtypes
• Fine-tuned transformers outperform prompt-based LLMs
• Explicit sarcasm labeling enhances detection accuracy