Probability-based spider monkey optimization-driven deep learning for intelligent fake news detection

Abstract

Fake news has more effects on spreading misinformation by reducing the scale and power of online social media, which degrades people’s trust in traditional journalism and the press and also changes the sentiments and opinions of the public. Therefore, this fake news detection is a crucial activity as it contains subtle differences between fake and real news. The major concept of this task is to promote a well-organized and reliable fake news detection method with intelligent technology. The pre-processing was initially performed and then it was subjected to the feature extraction phase. Here, the enhanced optimization algorithm termed as “Probability-based Spider Monkey Optimization (P-SMO)” is used for performing the selection of primary features. The detection of fake or real news is sophisticated by the Optimized Activation Function-based Deep Neural Network (OAF-DNN), in which the P-SMO helps to optimize the activation function to facilitate attaining better detection accuracy and high precision. From the overall evaluation of the results, the accuracy and precision of the offered model attain 98.3% and 98.65%. Experimental analysis of the offered approach is conducted by testing it with the baseline methods based on diverse evaluation metrics. Thus, the developed method outperformed the conventional methods to illustrate its superior performance. The developed fake news detection method can help to ultimately identify and debunk misinformation for generating optimal information, leading to enhanced public trust and decision-making. It has the ability to optimally detect unnecessary details in financial, online misinformation, healthcare, military, and election-related applications.