ASPECT BASED SENTIMENT ANALYSIS OF NEWS WEBSITES

Abstract

With the ever-increasing data on the internet especially regarding news, everything is becoming data intensive politics, business, sports, government, and other myriad of domains have been scoured for data by big companies over the last decade, World Economic forum has declared data a new class of economic asset, like currency or gold which makes it a resource to invest in. Text is a huge part of the data available online, this makes NLP (Natural Language Processing) a desired technique and is needed to face this mountain of unstructured data. News Websites publish thousands of articles each day, and some of them do put out harmful or even fake news on the internet (there has been a lot of work done regarding fake news) and to monitor these huge chunks of data it is impossible to be done by a single human or a team, and if we could do this in real time with machines this could save us tremendous time and resources. The primary purpose of our project is to develop a machine learning model that could be given a set of articles and it can produce the Aspect level sentiments of each article on a sentence base, we will have huge amount of data to be processed and labeled at every aspect, this is a building block for further implementation that can be an alert system, or a PR tool for analysis regarding an Entity.