A VictimFinder research led by Gear Lab member Mr. Bing Zhou, a first-year Ph.D. student, is recently published by the Computer, Environment and Urban Systems! Congratulations!
The manuscript is entitled "VictimFinder: Harvesting rescue requests in disaster response from social media with BERT," analyzed the global Twitter Social media which are playing increasingly critical roles in disaster response and rescue operations with advanced Natural Language Processing technologies. This study developed and compared ten VictimFinder models for identifying rescue request tweets, three based on milestone NLP algorithms and seven BERT-based. A total of 3191 manually labeled disaster-related tweets posted during 2017 Hurricane Harvey were used as the training and testing datasets. The performance of each model were evaluated by classification accuracy, computation cost, and model stability. Experiment results show that all BERT-based models have significantly increased the accuracy of categorizing rescue-related tweets. The best model for identifying rescue request tweets is a customized BERT-based model with a Convolutional Neural Network (CNN) classifier. Its F1-score is 0.919, which outperforms the baseline model by 10.6%. The developed models can promote social media use for rescue operations in future disaster events.
The online version of the manuscript is available on: https://www.sciencedirect.com/science/article/pii/S0198971522000680?dgcid=author
Figure: The architecture of NLP-based VictimFinder models.