Swamped with your writing assignments? Take the weight off your shoulder!
Below, there are train and test sets with archival data from Twitter. A binary label has been assigned next to each tweet indicating if it is describing/reporting or generally relevant to a world or domestic disaster. The goal here is to develop a monitoring system to turn twitter into a warning system for future disasters. To do this, however, a machine learning model would have to first distinguish humorous or figurative uses of words that could be consistent with a real disaster such as a forest fire.
Note that this is a topic in Data Science called Natural Language Processing, NLP