Exploring Sentiment Analysis: A Hybrid Approach using Deep Learning and Encoder

The rapid expansion of internet data underscores the increasing significance of timely analysis and monitoring of public sentiments online. In this context, analyzing sentiments related to public opinion events becomes paramount. Traditional word2vec models often fall short in fully capturing the depth of information within words. To address this, utilizing the BERT model for article feature extraction is proposed. This is coupled with a deep convolutional neural network to capture local article information. Subsequently, a fully connected network is employed to classify the articles, enabling effective sentiment analysis. Experimental findings from public datasets showcase the efficacy of sentiment analysis algorithms leveraging BERT in tandem with convolutional neural networks."