Chinese International Student Rental Reviews Sentiment Analysis Based on Python Crawler and Transformer Model

This study explores the complex process of sentiment analysis in reviews of international student rentals, a challenge made even more difficult by cultural, linguistic, and administrative barriers. The study utilized Python crawlers to collect and analyze 15,537 reviews from 30 popular study abroad cities. Neural networks and sentiment analysis techniques were employed to determine the sentiments of international students. To enhance the effectiveness of sentiment analysis, a Transformer-based model was implemented. After manual labeling and data cleaning, this model was developed and optimized through hyperparameter tuning, achieving high accuracy and F1 scores. Finally, we applied LDA topic modeling to extract nine topics and analyze each one by identifying its keywords, providing valuable insights into the rental experiences of international students in various urban environments.