一种基于BERT与依存句法的情感分析模型

Sentiment classification model based on BERT and dependency syntax

近年来基于深度学习模型的方面级情感分析方法已经成为了主流,特别是基于句法结构的图神经网络模型引起了研究者们的广泛关注。但大多数现有的模型对句法树的利用不够充分,无法准确地理解文本的语义。针对以上问题,提出了一种基于BERT与依存句法的情感分析模型。经过实验得出,对比于传统的机器学习方法及普通的深度学习方法,本文模型在准确率、召回率和F1值指标上均有明显提高。

In recent years,aspect‑level sentiment classification methods based on deep learning models have become mainstream,especially graph neural network models based on syntactic structures,which have attracted extensive attention from researchers.However,most existing models do not fully utilize syntactic trees and cannot accurately understand the semantics of the text.To address these issues,a sentiment classification model based on BERT and dependency syntax is proposed.Experimental results show that compared to traditional machine learning methods and ordinary deep learning methods,the proposed model achieves significant improvements in accuracy,recall,and F1‑score metrics.