Identifying Fake News On ISOT Data Using Stemming Method With A Subdomain Of AI Algorithms

Abstract

This investigation examines the adequacy of stemming strategies inside the subdomain of AI calculations for recognizing fake news on the ISOT dataset. Four calculations, to be specific Support Vector Machines (SVM), Naive Bayes, Recurrent Neural Networks (RNN), and Convolutional Neural Networks (CNN), were assessed through comprehensive tests. Stemming was connected to preprocessing the content information, upgrading the generalization capability of the models. Comes about show that CNN outflanked other calculations, accomplishing a precision of 90%, accuracy of 92%, review of 88%, and F1-score of 90% on the approval set. This signifies the suitability of advanced learning procedures in the implementation of complicated designs through literary material. In addition, the comparisons with related work represent a complete execution of our approach, the fundamental role of stemming strategies and AI calculations in fake news locations becomes clear. Creating future studies that are able to deploy the most sophisticated content preprocessing procedures and train learning strategies to enhance the efficiency and accuracy of fake news detection systems is advisable.