Detection of Fake News Using Heterogeneous Attributes
- 0 Collaborators
A news detection algorithm is designed to help users detect and filter potentially fake news varieties. The prediction of the potential for a particular story to be intentionally misleading is based on the analysis of true and false news. According to a study in the literature, among the current techniques, those that have the most possibilities of accuracy in the detection of false news are those based on machine learning and natural language processing. The general objective of this research is to analyze, compare and select techniques of machine learning and natural language processing that are fundamental in the development of fake news detection in order to construct an algorithm that can estimate the probability of the news being false and indicate the elements that contribute to the suspicion. ...learn more
Project status: Concept
Overview / Usage
The general objective of this research is to analyze, compare and select techniques of machine learning and natural language processing to define a methodology for extracting attributes from heterogeneous sources, creating algorithms for transforming attributes and adapting classification algorithms to deal with such attributes, a tool that, besides informing the probability of a news being false, be able to present an explanation of the reasons for the suspicion.
The objective can be divided into several tasks to be used as references of the progression of the work, being: to identify known sources of dissemination of false news; create a database containing fake news and real news; provide an up-to-date bibliographical survey of the state of the art; create a political bias indicator and propose a heterogeneous attribute extraction approach, which uses the most effective detection methods for the implementation of the fake news detection tool.