Publication

Abstract : Now-a-days social media has become a crucial part of life. Twitter is a social networking site on which people post and interact with messages renowned as tweets. Users registered officially can tweet, like and re-tweet messages. During the emergence of a disaster or crisis social media has become a significant means of communication. The widespread use of mobile phones and other forms of communication allows individuals to express and alert others about real-life disasters. Such knowledge relating to disasters spread over the media could save thousands of individuals by warning others and allowing them to take the required actions. It is being worked on by many firms to analyze tweets and observe tweets relating to disasters and emergencies using programming. Such efforts may be useful to loads of people using the internet. However, this effort has other problems, such as detecting and distinguishing catastrophe tweets from non-disaster tweets. Often the data available in twitter is not structured so processing is to be done on the data to classify data as ‘disaster’ and ‘non-disaster’. This paper deals with developing a model that can tell if a user is sharing data about a disaster. The data set used includes 10,000 tweets along with classifiers. This Optimized SVM model pre-processes the data using Natural Language Processing (NLP) and then builds the classifier model that gives maximum accuracy.