Twitter Sentiment Analysis using Apache Storm

Abstract

In today's highly developed world, every minute, people around the globe express themselves via various platforms on the Web. And in each minute, a huge amount of unstructured data is generated. This data is in the form of text which is gathered from forums and social media websites. Such data is termed as big data. User opinions are related to a wide range of topics like politics, latest gadgets and products. These opinions can be mined using various technologies and are of utmost importance to make predictions or for one-to-one consumer marketing since they directly convey the viewpoint of the masses. Here we propose to analyze the sentiments of Twitter users through their tweets in order to extract what they think. We classify their sentiments into three different polarities-"positive", "negative" and "neutral." Since, 6000 tweets are generated every second and this number is increasing, we need a robust system to process these tweets in real-time. Here, batch-processing would have its limitations and therefore a real-time and fault tolerant system, Apache Storm is used. After classifying the tweets, we represent the analysis in the form of graphs and charts which will enable our system users to understand public sentiments on the fly. This process as a whole is also called as Opinion Mining or voice of the customer.