Twitter sentiment analysis on fake news using python and natural language processing

Abstract

The aim of this thesis is to analyze the opinions, feelings, and emotions of Twitter users regarding fake news in account of the research process for the sentimental classification that is being conducted at the end. The theory has neglected to concentrate on fake news as a semantic analysis topic, something that this thesis addresses. It is asked in what level Twitter users express emphatical feelings and not only informative observations. This is done so the polarity of the public concerning fake news and misinformation can be determined, analyzed, and used in further researches regarding the detection of the aforementioned. Using Natural Language Processing (NLP) techniques, Python Programming Language, useful libraries as tweepy, pandas, NumPy, Textplob etc., and Jupyter Notebook as an integrated development environment (IDE), the final semantic score is determined. Every process and technique that participates to the calculation of the final score like data-extraction, data-preprocessing, data-analysis, and data-visualization are also part of the project and are displayed on it. It is shown that the majority of peoples' feelings toward fake news and misinformation in generally are more neutral than negative or positive. It is also shown that despite the numerical difference between the semantic categories, the more strongly expressed feeling was the negative one. The significance of this study is that it informs our understanding that most of the people doesn’t give much of their attention in the spread of fake news and when they do, most of the times they are strongly opposed to them.