Sixteenth International Conference on Technology, Knowledge & Society

Abstract

The interaction between news and social media can explain how social and political discourse has evolved with the rise of social media and people’s increasing reliance on it for news and information over mainstream news. This paper compares the nature of political discourses over social media and mainstream news media in India, and to determine whether discourse on social media is influenced by mainstream news reporting, using natural language processing. Two key current topics in Indian politics were identified for case studies; (1) the revocation of special status of Jammu and Kashmir, and (2) economic slowdown in India. News articles from four popular Indian English-language newspapers and Twitter posts (tweets) over a period of one year were collected. Based on past work done to understand news lifespan, it was determined that a particular news article may be discussed over a maximum period of 2 days on social media. Thus, news articles from a particular date are compared with Twitter discussions over the next two days of their respective publication dates, and were accordingly collated into respective documents. Using TF-IDF, Word2Vec, and Smooth Inverse Frequency to represent each document, a cosine similarity was calculated for the news document and corresponding tweets document. Our results indicate that very few discussions on social media reflect mainstream news reporting during that period, and that there may even be entire subtopics that are not covered over mainstream news but strongly influence social media.