Enhancing Multilingual Emotion Classification with Attention Mechanism for Transnational Education

In the context of transnational education, understanding emotional expressions across languages is essential for developing effective communication and fostering global perspectives in students. This paper uses innovative natural language processing techniques to explore sentiment dynamics in a multilingual setting. We propose an improved XLM-RoBERTa framework that integrates a multihead self-attention and attention pooling mechanism. The SemEval2018 Task 1 dataset is used as it covers Arabic, English, and Spanish. In particular, Low-rank Adaptation (LoRA), which is an attention mechanism, is used for efficient parameter fine-tuning, which made it scalable for deployment in resource-constrained educational environments. The results show that our proposed model achieves an accuracy of 0.87 and an F1-score of 0.85, higher than the baseline models. Finally, this research contributes to shaping innovative learning strategies that prepare students for a globally connected future.