ViGoEmotions: A Benchmark Dataset For Fine-grained Emotion Detection on Vietnamese Texts

TL;DR

This paper introduces ViGoEmotions, a large Vietnamese emotion dataset with 27 fine-grained emotions, and evaluates various transformer models and preprocessing strategies, highlighting the importance of emoji handling and annotation quality for emotion classification.

Contribution

The paper presents a new Vietnamese emotion dataset with detailed annotations and evaluates multiple transformer models, demonstrating the impact of preprocessing strategies on classification performance.

Findings

Converting emojis into text improves model performance.

Preserving emojis yields the best results for certain models.

Removing emojis decreases classification accuracy.

Abstract

Emotion classification plays a significant role in emotion prediction and harmful content detection. Recent advancements in NLP, particularly through large language models (LLMs), have greatly improved outcomes in this field. This study introduces ViGoEmotions -- a Vietnamese emotion corpus comprising 20,664 social media comments in which each comment is classified into 27 fine-grained distinct emotions. To evaluate the quality of the dataset and its impact on emotion classification, eight pre-trained Transformer-based models were evaluated under three preprocessing strategies: preserving original emojis with rule-based normalization, converting emojis into textual descriptions, and applying ViSoLex, a model-based lexical normalization system. Results show that converting emojis into text often improves the performance of several BERT-based baselines, while preserving emojis yields the best results for ViSoBERT and CafeBERT. In contrast, removing emojis generally leads to lower performance. ViSoBERT achieved the highest Macro F1-score of 61.50% and Weighted F1-score of 63.26%. Strong performance was also observed from CafeBERT and PhoBERT. These findings highlight that while the proposed corpus can support diverse architectures effectively, preprocessing strategies and annotation quality remain key factors influencing downstream performance.