A Blend Ensemble Classifier for Sentiment Decision Making in Indian Railways Tweets

Rakesh Kumar Donthi and Md. Tanwir Uddin Haider

Out of all social media platforms, Twitter is one of the Online Social Network that exhibits exponential growth of users and tweets every year. Tweets carry social feedback pertaining to different products or services catering to plenty of domains. In this paper, we present a framework for extracting sentiment on Indian railways tweets. Since, the result of any machine learning algorithm depends on the quality of the data set, so, in our framework, the first phase is the pre-processing model and we apply different filtering techniques. In the second phase, we clustered the tweets based on train number and features. Finally, we applied supervised machine learning using single and ensemble classifier with stacking on each cluster to extract the sentiment related on each tweet. The main objective of this paper is to make decision-support system of Indian railways based on features to passengers and organization. Our final results showed better sentiment classification compared to present state-of-art techniques.

Volume 11 | Issue 2

Pages: 145-162