Twitter Sentiment Visualization Using Deep Learning

Abstract

E-markets and public forums are major threads of generating and utilizing the crowd’s wisdom in terms of intents, behaviour and feedbacks, which are crucial for both buyers and sellers. Sentiment analysis is valuable extraction of user’s intents and experience through their online interactions on public forums and feedbacks. Automation of this task is very significant, especially to gauge commercial metrics about any product or service. Twitter being the most utilized social networking platform where the user gets to share his view about certain topic in a short text. Tweets can be utilized as source of data to detect the sentiment of the users over a topic. As capturing Subjectivity and representational ease trough discreet level of sentiment categories is challenging therefore this paper focus on effective classification though visual representation of sentiments. The proposed work uses an ensemble machine learning approach for sentiment classification of data and an extension of WORDCLOUD for sentiment visual representations. Through various experiments on real world data from twitter aptness and clarity of the proposed scheme can be easily observed.