Sentimental data analysis for prediction of public reaction using hadoop framework

Abstract

Sentiment analysis is taken into account to be a sub-class of machine learning and natural language processing. It's accustomed disencumber, identify, or depict opinions from completely diverse content structures, as well as news, reviews and editorials and sorts them as positive, neutral and negative. In this paper, we have an inclination towards investigating the effectiveness of linguistic possibilities for sensing the sentiment of Twitter messages. We have an inclination towards evaluating the usefulness of present lexical sources in addition to qualities that seize information regarding the natural and artistic language employed in microblogging. We take a administered attitude to the issue, however control current hashtags within the Twitter data for making training data. We are making use of Pig Latin in our system. We record the stream data and store it in .csv format file. Then we compare the words stored in file with AFINN dictionary and based upon the keywords provided, it will rate each keyword ranging from-5 to +5 depicting most negative to most positive comments. Those ratings are combined to obtain a numerical value and that is what gives us our prediction of public opinion.