Providing a comprehensive framework of effective features in detecting fake news: a systematic review

Over recent years, with the rapid development and increasing popularity of social media, we have seen a massive growth in the volume and variety of fake news. This phenomenon has profound effects on individuals and society. Verification is a widely used method to counter the negative effects of fake news. But this method is not efficient when analyzing huge amount of data. Therefore, advanced machine learning models and feature-based approaches are used to automatically identify fake news. At the same time, the large number of models and the heterogeneity of features used in the literature often create limitations for researchers trying to improve model performance. For this reason, in the present study, a comprehensive framework of the features used in the detection of fake news is presented with a systematic review method. In order to carry out this systematic review, using the guide provided by Okuli and Schabram, all studies conducted in the field of fake news using related keywords were taken from ScienceDirect, Springer, Emerald, IEEE, ACM, Wiley, Sage databases. , JSTOR, Taylor and WOS were extracted and finally 72 related articles were analyzed. As a result of the analysis of related articles, the features were placed in two main categories of news content and news context. News content includes linguistic and semantic features, visual features and style-based features. The news field also includes features based on user, post and network. The obtained results showed that the most used features in detecting fake news are features based on user profile, features of statistical style, writing pattern and readability. Due to the high variety of available features, it is suggested that a wide evaluation of features, models and their performance in multiple data sets should be done and in this way the performance of different models and feature sets should be compared in order to find the best combination of features in different conditions. to become clear.