Abstract
Sentiment analysis is one of the most crucial tasks in Natural Language Processing (NLP), involving the training of machine learning models to classify text based on the polarity of opinions. Pre-trained Language Models (PLMs) can be applied to downstream tasks through fine-tuning, eliminating the need to train the model from scratch. Specifically, PLMs have been employed for Sentiment Analysis, a process that involves detecting, analyzing, and extracting the polarity of text sentiments. Numerous models have been proposed to address this task, with pre-trained PhoBERT-V2 models standing out as the state-of-the-art language models for Vietnamese. The PhoBERT-V2 pre-training approach is based on RoBERTa, optimizing the BERT pre-training method for more robust performance. In this paper, we introduce a novel approach that combines PhoBERT-V2 and SentiWordnet for Sentiment Analysis of Vietnamese reviews. Our proposed model utilizes PhoBERT-V2 for Vietnamese, offering a robust optimization for the prominent BERT model in the context of Vietnamese language, and leverages SentiWordNet, a lexical resource explicitly designed to support sentiment classification applications. Experimental results on the VLSP 2016 and AIVIVN 2019 datasets demonstrate that our sentiment analysis system has achieved excellent performance in comparison to other models.
Abstract (translated)
情感分析是自然语言处理(NLP)中最关键的任务之一,涉及通过训练机器学习模型来根据意见的极性对文本进行分类。预训练的语言模型(PLM)可以通过微调应用于下游任务,从而无需从头开始重新训练模型。具体来说,这些PLMs已被用于情感分析过程,该过程包括检测、分析和提取文本情绪的极性。已经提出了多种模型来解决这一任务,其中基于RoBERTa优化了BERT预训练方法的PhoBERT-V2预训练方法脱颖而出,成为越南语最先进的语言模型。在这篇论文中,我们介绍了一种结合使用PhoBERT-V2和SentiWordnet进行越南评论情感分析的新颖方法。我们的提议模型利用了针对越南语进行了强大优化的PhoBERT-V2,并借鉴了专门为支持情感分类应用而设计的词典资源SentiWordnet。在VLSP 2016和AIVIVN 2019数据集上的实验结果表明,我们的情感分析系统与其他模型相比取得了卓越的成绩。 这一段文本概述了一个基于PhoBERT-V2和SentiWordnet的新情感分析方法,并强调了该系统的性能优势。
URL
https://arxiv.org/abs/2501.08758