Sentiment Analysis Based on the BERT and Bidirectional GRU Model

As the Internet continues to develop, a vast amount of user-generated textual data has emerged on various online platforms. Sentiment analysis has become a crucial tool for understanding user attitudes and emotions to analyze the specific emotional tendencies expressed in this data accurately. In this context, this paper employs a sentiment analysis model constructed using a BERT-based deep learning approach combined with a bidirectional GRU neural network. Initially, the BERT model is utilized to extract high-quality semantic features from the input textual data. These embedding representations extracted by the BERT model are then passed to the bidirectional GRU layer to capture sequential information further. After the GRU layer processes the features extracted by BERT, it ultimately determines the sentiment tendency. Experimental results on the IMDB movie review dataset demonstrate that the BERT-BiGRU prediction model significantly improves accuracy and F1 score compared to other models, providing an effective method and reference for the field of sentiment analysis.