Sentiment Analysis of Korean Modern Novel Texts Applying Deep Learning Models

Text Sentiment Analysis refers to the technology of analyzing text data and judging the sentiment tendency by using natural language processing and data mining. This paper focuses on the acquisition of textual sentiment in modern Korean novels and proposes a textual sentiment analysis model for modern Korean novels that combines the BERT-BGRU model with the attention mechanism. It utilizes the BERT pre-trained language model to replace the Word2Vec model for text vectorization representation, and uses the BGRU network for contextual novel information feature extraction. The experimental results show that the BERT-BGRU is 0.849 and 0.960 on the F1 value on SST-2 and IMDB datasets, respectively, and the BERT-BGRU methods all obtain better results. Taking Park Wan-su’s novel as an example, the research method is capable of better capturing the emotional changes of the original fictional text and illustrating the distance of its delicate emotions.