Prediction of Polarity in Online News Articles

Abstract: The importance of online news articles has evolved with the advancement of information and technology. How people gather information, shape their views, and engage with topics of relevance has been increased by the internet. Thus news articles become important sources and play a significant role in shaping personal and public opinion. Predicting polarity in news articles becomes crucial to have a well-balanced understanding of any event. Using aspect-based sentiment analysis, the application predicts sentiments attached to various aspects of a particular Hindi news article. Our approach consists of sentence identification using POS tagging techniques followed by aspect extraction using unsupervised learning algorithms and finally predicting the sentiments of the aspects using sentiment analysis. The predicted sentiments would be displayed in a user-friendly format so that the users can easily understand them. Existing systems work on the English language unlike our approach for sentiment analysis

Keywords: News Articles, Polarity, Sentiment Analysis, Topic Modeling, RNN