KumbhPulse: Echoes of Emotion from the Divine Gathering

This Dataset comprises reviews and comments on the Kumbh Mela. It contains 49,425 multilingual end-user-generated posts gathered from social media websites X (then Twitter) and YouTube, across Kumbh Mela festivities from 2013-2025, Except 2014, 2015, and 2020. Every record has sentiment-labeled Hindi, English, Marathi, and other regional Indian dialects content normalized to English using Google Translate API. The comments were scraped with Selenium For X (Formerly Twitter ) and YouTube API, and annotated with the VADER sentiment analysis tool into Positive, Neutral, or Negative classes. Engagement metrics such as likes and timestamps are included for temporal and behavioral analysis.

Data is saved in structured .CSV files divided by year and contain anonymized user IDs, content of comments, likes, post date and time, and sentiment labels. This dataset can be used for sentiment trend analysis, tracking of public opinion, and multilingual NLP uses during major cultural events. With its status as one of the first open datasets on the emotional trace of the Kumbh Mela, it is an essential asset for scholars of digital sociology, AI, sentiment computing, and public engagement analysis.