Abstract
Fake news can be understood as a version of an original piece of news with high dissemination capacity and the objective of deceiving, causing ambiguity or falsehood. It is mainly propagated through social media, its use being intentional and difficult to identify. In this context, researchers have been working on technological mechanisms to detect them. Thus, this research aims to analyze the accuracy obtained and the datasets used in fake news identification algorithms. This is an exploratory research with a qualitative approach, which makes use of a research strategy protocol to identify studies with the intention of analyzing them. As a result, we have the Stacking Method, Bidirectional Recurrent Neural Network (BiRNN) and Convolutional Neural Network (CNN) algorithms, with 99.94%, 99.82% and 99.80% accuracy, respectively. Kaggle, Weibo and FNC are the most frequently used datasets. It is suggested, for future research, to carry out research in addition to news about politics, the area that was the main motivator for the growth of research from 2017, and the use of hybrid methods for the process of identifying fake news.
Recommended Citation
Villela, Humberto Fernandes; Corrêa, Fábio; Nery Ribeiro, Jurema Suely de Araújo; Rabelo, Air; and Costa, Emerson Eustáquio, "Uma análise da acurácia obtida e datasets utilizados em algoritmos de identificação de fake news" (2022). ISLA 2022 Proceedings. 3.
https://aisel.aisnet.org/isla2022/3