Multi-Modal Emotion Labelling of Hinglish Utterances using Iterative Active Learning

The described method presents a system designed for labeling emotions in Hinglish utterances. It incorporates diverse types of information, encompassing text and emojis. The active learning approach is employed to iteratively refine emotion labeling by selecting challenging examples for human annotation. This method is particularly beneficial for understanding emotions in the linguistic and cultural context of Hinglish, which amalgamates Hindi and English elements. It holds practical applications in sentiment analysis and interpretation of customer feedback.