Comparison of Deep Learning and Random Forest for Rumor Identification in Social Networks

Abstract

The societal lifetime of each individual has created with online social media. These locations have made outrageous improvements in the socialize environment. The world’s targetable and fashionable Online Social Network (OSN) is Facebook, and it has brilliantly had more than a billion clients. It is a household to numerous kinds of antagonistic objects who misuse the sites by posting harmful or wrong messages. In few years, Twitter and other blogging sites have been around multimillion energetic users. It converted a novel means of rumor-spreading stage. The problem of detecting rumors is now more important, especially in OSNs. In this paper, we proposed rumor a different machine learning approaches as Naive Bayes, Decision tree, Deep learning and Random forest algorithm for identifying rumors. The experiment can be done with Rapid miner tool on everyday data from Facebook. The schemes of rumor identification are verified by smearing fifteen sorts based on user’s performances in Facebook data set to forecast whether a microblog post is a rumor or not. From the experiments, precision, recall, f-score value is calculated for all the four machine learning algorithms, further values are compared to find the accuracy (%) in all the four algorithms. And our experimental result shows that the overall average of precision for a Random forest provides 97% than the other comparative methods.