A review of methodologies for fake news analysis

Nowadays, with the proliferation of different news sources, fake news detection is becoming a crucial topic to research. Millions of articles are published daily in the press, on social media, and in electronic media, and many of them may be fake. It is common for scammers to spread fake news to mislead people for malicious purposes. For researchers to be able to evaluate fake news, it is necessary to understand its diversity, how to study it, how to detect it, and its limitations. A descriptive literature review has been conducted in this paper to identify more appropriate methodologies for analysing fake news. The review found two broad classifications in the fake news research methodologies: fake news study perspectives and fake news detection techniques. Based on our literature review, we suggest four perspectives to study fake news and two major approaches to detecting it. Fake news can be studied in terms of knowledge, style, propagation and source. In order to detect fake news, there are two major approaches: manually and automatically. There are two types of manual fact-checks: expert-based and crowd-sourced. Automatic techniques are based mainly on data science methods, specifically deep learning and machine learning. A machine learning-based method was found to be more appealing when we evaluated all the automatic methods. Further research will focus on investigating the efficacy of using Bayesian methods for detecting fake news statistically because it is a flexible approach that allows for rapid updating of models in response to new data and has been successfully applied to a wide range of problems across different domains.