MY-BERT System: Breaking Language Barriers with AI-Powered Sentiment Analysis

About the Project

The growing trend of code-mixed language in online communication poses a challenge for traditional sentiment analysis systems that may have difficulty in accurately understanding multilingual and mixed-language expressions. The purpose of this project is to conduct sentiment analysis on social media data, particularly YouTube comments from entertainment topics, which is Malay-English code-mixed language namely “Manglish”. For this purpose, a code-mixed prototype of sentiment analysis was designed based on the BERT model. All of this was achieved via Python, Hugging Face Transformers for BERT fine-tuning and the YouTube Data API for the data, and Streamlit for the prototype being created in a web app.