Authors: Xiaohui Liu*, The University of Southern Mississippi, Bandana Kar, Oak Ridge National Lab
Topics: Hazards, Risks, and Disasters, Geographic Information Science and Systems, Spatial Analysis & Modeling
Keywords: Relevance, Reliability, Twitter, Spatiotemporal Data Mining
Session Type: Paper
Start / End Time: 10:00 AM / 11:40 AM
Room: Napoleon D1, Sheraton 3rd Floor
Presentation File: No File Uploaded
While Twitter has been touted to provide up-to-date information about hazard events, the reliability of tweets is yet to be tested. This research examined the reliability aspect of risk information extracted from Twitter during the 2013 Colorado floods using two different approaches. The first approach developed a classifier to categorize the tweets using a machine learning algorithm, then link different external sources to validate the “trueness” of each tweet. The validation uses algorithm and criteria including data generator, completeness, accuracy, detailedness, and others to rank the tweets based on “trueness”. The second approach employs Google Cloud Vision to detect features related to floods and damages in images to automatically extract risk information. In comparison, manual identification of features is used to validate the accuracy. Both approaches work in parallel to ensure the extraction of reliable risk information to increase situational awareness. The use of Google Cloud Vision provides an efficient tool to automatically extract useful risk information during emergencies.