raw-tweets, multi-class or binary classification.
{
"text":"I was wondering why the room was shaking when I was trying to sleep ... I found out it was an earthquake. I bring disasters with me"
"label":"direct-eyewitness"
}
{
"text":"F*cking hell...my wife and kids are in Tokyo and they're in the middle of an earthquake Jesus Murphy just how crap can one day get?"
"label":"indirect-eyewitness"
}
{
"text":"It was 90 degrees at noon and now we have a thunder storm and flash flood warning.... welcome to SoCal"
"label":"vulnerable-direct witness"
}
...@article{zahra2020automatic,
title={Automatic identification of eyewitness messages on twitter during disasters},
author={Zahra, Kiran and Imran, Muhammad and Ostermann, Frank O},
journal={Information processing \& management},
volume={57},
number={1},
pages={102107},
year={2020},
publisher={Elsevier}
}