Integrated Approach for Sentiment Detection and Emotion Recognition in Code-Mixed Hinglish Data

With the growing prevalence of social media world-wide,Analysis of emotions and feelings in multilingual and hybrid texts explores how feelings are expressed in mixed-language contexts has gained considerable importance. While there have been notable advancements in research focused on the English language, there is still a lack of comprehensive studies addressing code-mixed languages, particularly those that blend Hindi and English. To improve existing methodologies, we propose enhancements to a multitask framework designed for detecting sentiment and emotions within code-mixed Hinglish data. Our approach aims to increase model accuracy of 0.66 % by expanding the training dataset, refining the testing and validation datasets through improved annotation, and incorporating multilingual datasets beyond Hindi-English. By leveraging a larger and more diverse dataset, our proposed model seeks to surpass current performance benchmarks in multilingual sentiment and emotion analysis. contributing valuable resources for future research.