Smart Disaster Management Using Big Data Analytics
Sustainable Aviation, Springer Nature, ss.175-179, 2023
- Yayın Türü: Kitapta Bölüm / Araştırma Kitabı
- Basım Tarihi: 2023
- Doi Numarası: 10.1007/978-3-031-37160-8_27
- Yayınevi: Springer Nature
- Sayfa Sayıları: ss.175-179
- Anahtar Kelimeler: Big data, Disaster analysis, Natural language processing, Smart city, Text classification
- Galatasaray Üniversitesi Adresli: Evet
Özet
The direct or indirect affection of the disaster is a severe issue in the analysis of smart cities. The behavior of public information is vast, and the detection of victims and potential risks is time limited. Social networks provide live information within the disaster region where the emergency and rescue organizations would reach the critical zones. However, the disaster knowledge with critical insights is generally flooded with non-rescue information which is overwhelmed through different modalities in big data analytics. Therefore, the guidance of big data represents the foundations in smart disaster management. In this study, we have focused on the 2020 Izmir earthquake which is classified as a severe earthquake in the intensity scale. The earthquake information has been retrieved using microblogs from Twitter. The dataset has been preprocessed manually and automatically. The manual labels have been trained as vector embeddings in order to generate automatic labels as a semi-supervised approach. Naive Bayes, support vector machines, and BERT transformer networks have been applied on two classes. All approaches scored relevant evaluation values for disaster knowledge. Our findings presented the efficacy of big data approaches for rescue and non-rescue classes in disaster management. We conclude that smart rescue strategies would rely on big data analytics where the civil rescue teams outnumber the emergency and rescue organizations.