Ukrainian Resilience: A Dataset for Detection of Help-Seeking Signals Amidst the Chaos of War

The "Ukrainian Resilience" dataset is a novel collection of Ukrainian language social media posts from the ongoing Russia-Ukraine war, designed to facilitate the development of natural language processing and machine learning models for detecting subtle linguistic cues indicating distress or urgent requests for aid. Comprising a binary classification of posts requiring help or not during the conflict, this dataset aims to enhance humanitarian efforts by enabling more rapid identification of genuine calls for assistance amidst vast online content. Leveraging advanced language models and algorithms to discern nuanced signals of distress, the "Ukrainian Resilience" dataset holds significant potential for real-world impact, as evidenced by preliminary experiments using GPT-3.5 achieving 81.15% accuracy.