Intelligent based Framework for Detection of Fake News in Social Media Platforms

Abstract

Developing a framework for the detection of fake news that is based on a conceptual and intelligent framework will improve the detection of fake news in the social network platforms. It has been hypothesized that if not checked, the increasing spread of fake news may either be the immediate or remote cause of third world war. The existing frameworks suffer in two fundamentals of social concepts and artificial intelligence based revolutionary trends. The focus of this research work is to propose a fake news detection framework that combines the structurally modeled fake news concepts and artificial neural network model. The study used a cross-sectional model testing correlational design which formed the basis of developing the intelligent framework. A simple random sampling technique was used to determine the sample size. The research instrument used was questionnaire administered through Google form survey. The instrument was validated using content, construct and criterion validity. During the pilot test, the reliability of the instrument was established with the standard value of ≥ 0.7 as the benchmark. The method of data analysis used was Pearson Product Moment Correlation Statistics within a significant level of 0.05. Analysis of moment structure tool was used to answer the research hypothesis using structural equation modeling technique. The data obtained from the questionnaires and the test were analyzed using the exploratory factor analysis process (a first generation statistical method of analysis), and the expected designed model, based on Structural Equation Modelling was evaluated using standard goodness of fit indices (GOF) of confirmatory factor analysis (a second generation statistical method of analysis). The dataset generated, measured, analyzed and modeled formed the basis for the development of intelligent based framework. The proposed artificial neural network framework comparatively achieved better detection rates over the conventional and structurally modeled datasets. The framework addresses the social disposition and artificial intelligence based fake news detection gaps.