Sentiment Analysis in English-Punjabi Mixed Social Media Posts

This paper presents a model that detects emotions in English-Punjabi code-mixed social media posts. The pervasiveness of multilingual code-mixed communication on social media platforms poses a significant challenge for sentiment and emotion analysis. Our work addresses this gap by developing a logistic regression-based emotion detection system that classifies posts into three divisions: positive, negative, and neutral. The data set used for this study comprises 37,805 training posts and 10,307 test posts, each labeled with an emotion. The model was trained and evaluated after preprocessing the code-mixed text data, including cleaning, stopword removal, and vectorization. The model attained an accuracy of 92% on the test dataset, exhibiting its effectiveness in handling code-mixed content. Furthermore, we implemented a real-time web interface using Streamlit, which allows users to classify the emotions of mixed English-Punjabi code posts. The promising results suggest that this approach can be beneficial in various implementations such as mental health tracking and sentiment analysis in multilingual contexts.