Social Media Text Sentiment Analysis: Exploration Of Machine Learning Methods

With the increasing number of social media and online platforms, sentiment analysis has become an important research object in new fields. This study aims to explore and improve text sentiment analysis methods to improve the ability to extract and understand sentiment information in social media data. A public data set called “Sentiment analysis book review on Amazon Kindle” is used, which contains product analyses in the Amazon Kindle store category, with a total of 982,619 records. Through effective data preprocessing, including text cleaning, tokenization, removal of stop words and lemmatization, we are ready for the subsequent development of sentiment analysis models. This study uses natural language processing technology to classify reviews into positive and negative categories, and uses RNN and logistic regression machine learning algorithms for sentiment analysis and comparison. Finally, a basic web application prototype based on FLASK, HTML, CSS and Python was designed, integrating AI models to provide sentiment analysis results and data visualization in real time.