Isolating rumors using sentiment analysis

Abstract

In recent days, social media has become a platform to spread false facts all the way through internet. One of the growing data analytic engine from web informal organization says, twitter has become the prime source for spreading fake news facilitating numerous perpetrators around the globe. It has turned into a competent, speedy cum effortless hotspot for news-fans to just click and forward junk data. Individuals are opting to use twitter for searching information regarding crisis circumstances and everyday occasions. In twitter, the spread of fraudulent or inaccurate information during the emergency situations will affect the individuals and public in numerous ways, but the original news is more reliable when it is declared by the news channels. So, the importance of proposed framework is categorized in three steps; Initially, Twitter Scraper is applied to scrape the vast volume of tweets and metadata from the collected set of tweets for the study on former Chief Minister (state of Tamil Nadu) death case controversy during 2016. Then the threshold value based on negative polarity of common tweets for the scraped data is calculated, once the tweeted texts are different from the threshold condition it will be automatically tagged as 'rumor' (negative) or else tagged as 'non-rumor' (positive) using sentiment classifier. Finally, the proposed model on VADER based sentiment analysis identifies the false facts. It is obtained as a result of the sample tweets regularly training and testing it on whole datasets.