Sentiment Analysis in Disaster Management using Tweets

Social media platform such as Twitter is becoming a source for event-based early warning systems. These event-based tweets act as input to crisis management organizations for decision making. In this paper dataset containing tweets of the sandy hurricane is used for sentiment analysis. Sentiment analysis is used to analyze tweets as positive, negative or neutral. These tweets are manually annotated with sentiment labels using Sentiment analysis API which uses Long Short Term Memory algorithm (LSTM).Naive Bayes Multinomial Text and the zeroR classifier is applied using Weka on the trained dataset. The accuracies of two classifiers are examined and compared. The result shows that Naive Bayes Multinomial classifiers show the better result when use training set test option is used with the balanced dataset.