Automated Data Analytics in Measuring Brand Sentiment Using Ml Technique from Online Conversations—Twitter

Abstract

With the ever growing amounts of data there is good motive to believe that smart data analysis will become more pervasive and an obligatory component in the progress of industry. Social Media act as a source between the company and the customers. People being at their work place can get to know the current status, discussions done by people on any trends and even he/she can express their opinions and views as per their interests through blogs, forums, emails etc. Twitter being a popular micro blog allows people to chat about their words with public in form of short texts (140 characters). Researchers of NLP and DM are attracted towards the sentiment analysis from past few years due to its many tricky research problems and purpose. In this paper, a novel Machine Learning Approach is used to classify the twitter dataset. The working of the Algorithm is explained with Sample set taken from twitter. If we have very little information i.e. the training sample, Algorithm defined may not give accurate estimation/probability of any object belonging to particular class as we have no other information to obtain a better estimation. This estimate will be reasonable if a training sample is very large and properly chosen. Thus in this paper Classification helps us to find the belongingness of an instance to a class to find the collective opinions of the users of twitter.