Phailin tweets

What makes this dataset particularly valuable is its multi-layered structure. Beyond the raw tweets, we have developed a processed user-level dataset with derived metrics such as spread_score and focus_rate that researchers can use to analyze information diffusion patterns.

Additionally, we include 1,500 manually annotated tweets with labels for user types and content categories, which we used to train a few-shot classification model achieving a recall score of 0.72. This annotated subset can serve as a benchmark for future machine learning applications in disaster communication research. The dataset addresses a significant gap in available resources for studying disaster response in the Global South context.