Sentiment Analysis of Pulwama Attack Using Twitter Data

Abstract

Microblogging sites like Twitter have become significant sources of real-time information during a disaster. Millions of tweets are Microblogging sites like Twitter have become significant sources of real-time information during a disaster. Millions of tweets are posted during disasters. This algorithm is applied to twitter tweets to extract the sentiments of the public on such types of manmade disasters. In order to use microblogging sites effectively during disaster events, it is needed to summarize the large amounts of real-time non-situation information posted on twitter. In this study, non-situational tweets were analyzed which were posted during the recent disaster event of the Pulmawa attack. The proposed methodology is to develop a Gradient Boosting classifier using machine learning techniques to achieve better performance compared to support vector machine and random forest classifier to categorize various types of non-situation tweets collected during disaster into a set of different classes. Well-known sentiments were used for mining that exhibits eight basic emotions, that is, joy, Trust, fear, surprise, sadness, disgust, anger, and anticipation.