Document Type : Research Paper

Authors

• Mohammad Mohsen Sadr ¹
• Afshin Mousavi Chelak ¹
• Soraya Ziaei ¹
• Jafar Tanha ²

¹ Faculty Member of Payame Noor University, Tehran Iran.

² Electrical and computer engineering department, University of Tabriz, Tabriz, Iran.

Abstract

False rumors and news are always published as purposeful approaches with social, economic, political intents in order to provide false information and deceive people in the communities. This leads to a lack of trust in news and information. Differentiating real news from rumor has been considered as one of the most important aspects of news evaluation and different approaches have been used to identify and distinguish fake news from real one. Among them, the use of artificial intelligence and machine learning methods has been more important due to the successes achieved. Due to this advantage, the present study has attempted to use machine learning algorithms including SVM, k-NN, decision tree, random forest and MLP, to identify and classify fake and real news in the data set collected from Persian Twitter messenger. Based on the results of the confusion matrix implementation and functional evaluation of learning algorithms, it has been determined that Randomized decision trees and decision tree have the highest accuracy in evaluations with 90.25 and 90.20 as in the next step, the accuracy of the random forest is 89.99\%. This indicates the ability of tree decision-making algorithms in optimal evaluation and better identification of fake news on Persian Twitter. Also, random forest and Randomized decision trees algorithms have the highest precision in implementation with 92%, and after these two algorithms, decision tree with 90.20% is in the third rank of precision.

Keywords

• Fake Persian News
• Artificial Intelligence
• Machine Learning
• Learning Algorithms
• Twitter Messenger