Emotion Recognition and Big Data Analysis using Bert: Taking Natural Language Processing as an Example

With the rapid development of social media and online communication, emotion recognition, as an important natural language processing technology, has demonstrated its application value in many fields. Although traditional rule-based methods play a pioneering role, their universality and accuracy are often limited. This paper aims to explore the combination of emotion recognition and big data analysis, and uses multiple natural language processing methods to conduct emotion analysis on large amounts of text data. Our approach includes deep learning-based emotion classification algorithms, as well as the optimization and improvement of the emotion dictionary. The experimental results show that the highest recognition accuracy is $95.4 \%$ when using the modified model for emotion recognition. This result not only validates the effectiveness of the proposed method on real datasets, but also provides new ideas for future studies of sentiment analysis. The conclusion is that the continuous development of emotion recognition technology will greatly promote human-computer interaction, market analysis and social media monitoring.