The largest publicly available, human-annotated, inter-annotator-agreement-validated emotion dataset for Roman Urdu.
RUEmoCorp (Roman Urdu Emotion Corpus) is a large-scale, manually curated, expert-annotated dataset of Roman Urdu social media and conversational texts labeled across 7 emotion categories: joy, anger, sadness, fear, disgust, surprise, and none. It is the training corpus behind roman-urdu-emotion-xlmr-v2 — the… See the full description on the dataset page:
https://huggingface.co/datasets/Khubaib01/RUEmoCorp.
