Approximations to geolocation of disaster related tweets

• Main Author: ROSALES, JOHN CLIFFORD
• Format: text
• Published: Archīum Ateneo 2017
• Subjects: Twitter; Communication in crisis management; Emergency management > Geographic information systems; Emergency management > Data processing; Geographic information systems; Online social networks
• Online Access: https://archium.ateneo.edu/theses-dissertations/182; http://rizalls.lib.admu.edu.ph/#section=resource&resourceid=1265467000&currentIndex=0&view=fullDetailsDetailsTab
• Institution: Ateneo De Manila University

The use of tweets as information aid during disasters has been limited by the lack of location information in majority of the tweets. This study created two algorithms to approximate tweet location based on the text content of the tweets. The first algorithm used machine learning algorithms to predict the distance of a tweet from the eye of the typhoon and the disaster affected area. The second algorithm employed semantic modeling and comparison to predict the location of a tweet as latitude-longitude coordinate. The results of these studies show that temporal factors are important in creating more accurate location approximation models. Models that predict a tweet's relative distance to the affected area have also been shown to be more effective than models that predict relative distance to the eye of the typhoon.