Efficient Feature Extraction with Optimization Based Hybrid Classification Model for Fake News Detection

Abstract

Abstract: A massive volume of information is circulated daily over online and print media; however, it is difficult to say whether the data contains truth or is fake. Due to their lower costs, simple accesses, and speedy distributions, social media have recently become one of the primary worldwide news sources. Dubious News also have a cost involved even though reliability and sizable possibility of reading "fake news" that was purposefully released to misinform readers [1,2].  Social media has grown in popularity as a means of consuming news over the last ten years because it is simple to use, spreads information quickly, and is inexpensive. Hence, fake news detections from social media have gained traction in research studies. Social media introduces new characteristics and issues for detecting false news, rendering traditional news industry's existing detection algorithms ineffective or irrelevant [3]. Recent research compares fake and authentic news strips using critical keywords in hybrid classification algorithms. The three steps described in this study can assist in identifying false news from social media information. The unstructured data from the media are transformed into structured form using a number of pre-processes. The Lexicon Model extracts the features and unknowable properties of misleading news in the second stage. This research project is currently in its third stage. a feature selection method by WOA (whale optimization Algorithm) for weight value to tune the classification part. Finally, a hybrid classification model called WOA-HYC, a hybrid fuzzy based CNNs (Convolution Neural Networks) and kernel based SVM (Support Vector Machine) to identifies false information and its effectiveness is measured in terms of values obtained for accuracy, precision, recall and F1 score. Experiment results show the suggested model achieves higher results amongst methods considered in this study.