Proceedings of The 11th International Conference on Research in Engineering, Science & Technology
Year: 2021
DOI: https://www.doi.org/10.33422/11th.restconf.2021.03.15
Structure and Experiments to Recognition and Classification of Fake News about Citizens in Brazilian Government Positions
Igor Baesse de Araujo; Yuri Rezende Mustifaga, Luiz Melk de Carvalho, Diva de Souza e Silva Rodrigues, Flávio Henrique Batista de Souza
ABSTRACT:
The high flow of information generated an effect, where the news started to have characteristics that disprove their reliability index, generating a problem called fake news. Thus, a validation methodology was created using NLP (Natural Language Processing) and SVM (Support Vector Machines) pattern recognition algorithms for investigating news with accuracy measured through the analysis of AUCs (Areas Under the Curve – varying between 0 and 1) and represented via Chatbot. With a methodology focused on experimentation, firstly, for the collection of information and news, a webtool was developed for the gathering and synthetic analysis of the main websites that publish news in Brazil, such as web crawler. An indexing function was developed for all website addresses, correlated to a research factor (which in this work is Brazilian citizens in government positions). The web crawler’s distribution network resembles a graph, where the vertices are the ten websites most relevant to the public and the edges are the interconnections between the websites. For the recognition of fake news patterns, during the website analysis, a list of words for research is generated, and the SVM algorithm performs a non-linear separation between fake news and true news. To disseminate the results of the fake news analysis to citizens, a chatbot was developed. In this chatbot the user sends a snippet of the news or the web address of the news with some keywords and the tool will handle the processing of the request. The assertiveness measured by the AUC between 0.96 and 0.99.
Keywords: Fake news; Natural Language Processing; Support Vector Machines; Chatbot; Citizens in government positions.