Navegando por Assunto "Arquitetura de Big data"

Agora exibindo 1 - 1 de 1

Acesso aberto (Open Access)
Uma Proposta de arquitetura de big data para detecção de fake news.
(Universidade Federal do Pará, 2020-01-24) QUEIROZ, Daniele Moura de; FRANCÊS, Carlos Renato Lisboa; http://lattes.cnpq.br/7458287841862567
In last years, a large amount of information has been transmitted through the Internet, especially in social media, providing greater knowledge on various topics, but making people susceptible to false information that can cause various damage. Although it is not a recent phenomenon, the sharing of false news has been a matter of concern for specialists and the population in general, since it can cause impacts of national and even global proportions. The transmission of fake news can cause various damages, from financial to losses related to defamation, injury, offense, reputation or dignity of people or organizations. The spread of this false information has made it difficult to detect reliable news sources, increasing the need for computational tools that can help identify the reliability of digital content. Moreover, the massive amount of data generated daily at high speed and different types of formats such as text, images, videos and audios, makes analysing this data a big challenge. With the advent of big data technologies, it is possible to use a range of tools and techniques to efficiently store, process and analyse massive data to help investigate the credibility of news disseminated and shared by middle of the internet. This paper discusses the importance of Big Data to combat fake news, based on an appropriate conceptual and technological framework, and presents a Big Data architecture proposal for storing, processing and analysing large data sets, aiming to assist in the investigation of truth of news. For this, experiments were performed using a mass of data containing different formats, i.e. structured and unstructured data, extracted from news sources and forming a corpus composed of false and true news. This mass of data was stored in a Hadoop cluster using the Hadoop Distributed File System (HDFS). The corpus was processed using the MapReduce programming model and the news classification was performed using the Naive Bayes algorithm from Mahout library, obtaining an accuracy of 99.74%. The preliminary results produced by the development of this study reveal an architecture capable of storing, processing and analysing Big Data in the context of fighting fake news.