Identification of Disaster-Related Tweets Using Natural Language Processing: International Conference on Recent Trends in Artificial Intelligence, IOT, Smart Cities & Applications (ICAISC-2020)

Abstract

In today’s world social media has become an integral part of life. Twitter is an American micro-blogging and social networking portal which provides the users a platform to post news, data and thoughts. Twitter has become an essential mode of communication medium during the occurrence of an emergency or disaster. The pervasiveness of smart phones and tablets enable people to pronounce and inform others of the occurrence of an emergency they are experiencing in real-time. This information regarding disasters propagated over the social media can save thousands of life by alerting others so that they can take evasive action. Many agencies are trying to programmatically analyze tweets and recognize disasters and emergencies. This kind of work can be beneficial to millions of people connected to the internet, who can be alerted in the case of an emergencies or disaster. But the real challenge lies in segregating the tweets trying to announce a disaster from the ones which are not related to a disaster. Twitter data is unstructured data, thus Natural Language Processing (NLP) has to be performed on the twitter data to classify them into classes –“Related to Disaster” and “Not related to Disaster”. The paper does a prediction on the test set created from the original data-set. It does an accuracy testing of the classifier model generated. This paper uses decision tree classification mechanism for building the classifier model and for making predictions.