Sentiment Analysis of Mandarin Social Media Text Using an Adjective-Enhanced Multi-Head Attention Model

Authors

  • Jie Zhang Author
  • Ruhaila Maskat Author
  • Zhiping Zhang Author

Abstract

The advent of research into neural networks has led to the widespread utilisation of deep learning methodologies in the domain of text sentiment analysis. In recent years, Transformer and its variant Bert have attracted considerable attention, and their performance has been demonstrated to be superior in practice. However, it is important to note that Mandarin’s unique social media dataset is characterised by a variety of languages, distinct cultural backgrounds, and intricate emotional expressions.The significance of word order is inherently apparent. Meanwhile, the potential of adjectives in emotional expression is frequently disregarded. The conventional Transformer model's efficacy in feature extraction is constrained in certain scenarios, impeding the efficiency of its implementation. It is necessary to adapt and optimize the model further to account for the specific attributes of Mandarin social media data in the task design.We propose a novel Transformer-based sentiment classification model, Adjective Enhanced Multi-Head Attention (AE-MHA), which incorporates the weighted significance of adjectives into the self-attention mechanism. The model is trained and evaluated on three Chinese social media datasets.The result show that the proposed model can extract richer text features than other baseline models, and can achieve better classification results than other baseline models.These findings suggest that AE-MHA is particularly suited to the linguistic nuances of Chinese social media text, providing a robust framework for future research and applications in Mandarin sentiment analysis.

Downloads

Published

2026-03-25

Issue

Section

Articles