Empowering Twitter Sentiment Analysis With BERT: Exploring the Role of Pre-Training Data

Social networks are widely used, and as a result, a tonne of user-generated data is readily accessible to the general public. This data can be analyzed to understand more about people's perspectives. In this study, we analyzed the use of BERT, a bidirectional encoder representation from transformers, for the sentiment analysis of a million tweets from different Twitter users. We constructed a classifier and used the real-world dataset of tweets to assess the effectiveness of the model that was created for the purpose of sentiment analysis of tweets. The studies conducted have shown that the modified model was able to acquire an accuracy rating of 0.97 for sentiment analysis.