Mapping of Tweet Location with Sentiment Analysis (SMTL)

Abstract

Accumulating public view by exploring social media data has appealed to researchers due to its wide impact. The people behavior or reaction on certain events can be tracked by use of various text mining methods. In this paper, two new approaches for sentiment analysis for twitter data tidyverse and deep learning are proposed. The proposed methods are analyzed for two of Government of India decisions demonetization and removal of Article 370. The tweets are classified as positive, negative and neutral by calibrating scale for sentiment score. The glmnet classifier is used in deep learning approach for training the twitter text data. The proposed methods give promising results of people opinion about these two government decisions. The preprocessing considers TF-IDF features for modeling glmnet classifier. The user location of twitter users is mapped with use of GIS tools which shows overall impact of these events within the country and world. The results of proposed method show improved accuracy in determining positive and negative views of people.