Özet:

In natural disaster management, the damages caused by natural disasters can be minimized by using Geographic Information Systems (GIS) in pre-disaster preparation, disaster response and post-disaster recovery stages. The aim of this study is to scrape X (formerly known as Twitter) social media data related to the Kahramanmaraş earthquakes on February 6, 2023 using Selenium and BeautifulSoup libraries in Python programming language, and to examine the post-disaster emotional states of people affected by the earthquake using natural language processing and deep learning methods. Thus, it will be possible to contribute to the planning of the general emotional state of the region at the time of the earthquake and the social and psychological rehabilitation activities to be carried out afterwards in a faster and easier way in the GIS environment. In this study, a sentiment analysis was performed with 87% test accuracy on the scraped and organized dataset from the X platform using the Gated Recurrent Units (GRU) deep network model in natural language processing. In addition, by performing hot spot analysis in the GIS environment, the clustering pattern of the emotional state in X messages related to the earthquake occurred was statistically analyzed. Thus, it was determined that georeferenced social media data of the X platform related to possible earthquakes can be used in sentiment analysis.

Anahtar Kelimeler: