Fake news detection in social media based on sentiment analysis using classifier techniques

Fake news on social media, has spread for personal or societal gain. Detecting fake news is a multi-step procedure that entails analysing the content of the news to assess its trustworthiness. The article has proposed a new solution for fake news detection which incorporates sentiment as an important feature to improve the accuracy with two different data sets of ISOT and LIAR. The key feature words with content's propensity scores of the opinions are developed based on sentiment analysis using a lexicon-based scoring algorithm. Further, the study proposed a multiple imputation strategy which integrated Multiple Imputation Chain Equation (MICE) to handle multivariate missing variables in social media or news data from the collected dataset. Consequently, to extract the effective features from the text, Term Frequency and Inverse Document Frequency (TF-IDF) are introduced to determine the long-term features with the weighted matrix. The correlation of missing data variables and useful data features are classified based on Naïve Bayes, passive-aggressive and Deep Neural Network (DNN) classifiers. The findings of this research described that the overall calculation of the proposed method was obtained with an accuracy of 99.8% for the detection of fake news with the evaluation of various statements such as barely true, half true, true, mostly true and false from the dataset. Finally, the performance of the proposed method is compared with the existing methods in which the proposed method results in better efficiency.