Analyzing Twitter Sentiments Through Big Data Analytics

Abstract

The advent of social media has generated a lot of buzz among Internet users these days. A number of social networking sites are used these days, which has led to the rise of sentiment analysis. Twitter is a popular site where users post comments in the form of short status messages. Millions of tweets are received every year and sentiment analysis of these tweets interest among Internet users today. Data from these social networking sites can be used for a number of purposes, like prediction, marketing or sentiment analysis. Twitter is a highly used social media site for posting comments through short status messages. The millions of tweets received every year could be subjected to sentiment analysis. But handling such a huge amount of unstructured data is a tedious task to take up. The current Analytics tools and models used that are available in the market are not sufficient to manage big data. In this thesis, we make use of Apache Mahout along with its Hadoop functionalities to carry out sentiment analysis Hadoop is a framework that performs computations over large datasets. With its framework MapReduce, it divides queries among different nodes nodes with computations to be performed in parallel. This provides faster query execution and faster result provision. In this thesis, we take up the opinions of people on the services of Airtel. These opinions are converted into a training set and Mahout is used to carry out Naïve Bayes classification to decide how many tweets are correctly classified into being positive, negative and neutral.