Published Paper Details:

In last decade of time, people are heavily relied on social media environments like Twitter, Facebook, Blogs, Whatsapp, Instagram etc., to know about the news about different concepts, famous people, products and services. Most of the people utilized these environments for sharing their opinions on different entities. Some people are spreading false or fake information in these environments to misguide the users. Identification of people who spreads the fake information becomes one important challenge for research community. PAN competition conducted a competition on fake news spreaders detection task in 2020. Several researchers proposed solutions for finding the authors who spreads the fake news in social media environments. In this work, we proposed an approach for fake news spreaders detection by using different feature selection algorithms. The content based features like words are most important features to differentiate the writing styles of fake news spreaders and real news spreaders. The identification of important words for experimentation is very important to improve the accuracy of fake news spreaders detection. In the proposed approach, feature selection algorithms are used for identifying most relevant words or features for experimentation. The identified features are used for representing the documents as vectors. These document vectors are trained with machine learning algorithms for generating the classification model. This model is used for predicting the accuracy of proposed approach as well as for predicting whether new author is fake news spreader or real news spreader. The PAN 2020 competition fake news spreader detection dataset is used in this experiment. Two machine leaning algorithms such as random forest and support vector machine are evaluating the accuracy of proposed approach. The proposed approach attained best accuracy for fake news spreader detection when compared with most of the approaches.