Full text
| |
| Author(s): |
Émerson Yoshiaki Okano
Total Authors: 1
|
| Document type: | Master's Dissertation |
| Press: | Ribeirão Preto. |
| Institution: | Universidade de São Paulo (USP). Faculdade de Filosofia, Ciências e Letras de Ribeirão Preto (PCARP/BC) |
| Defense date: | 2020-03-13 |
| Examining board members: |
Evandro Eduardo Seron Ruiz;
Evandro Marcos Saidel Ribeiro;
Oto Araujo Vale
|
| Advisor: | Evandro Eduardo Seron Ruiz |
| Abstract | |
The web is an environment where people post and search any type of information on the most diverse topics. However, the information found on the web is not always truthful. There are malicious users who post deceptive information intending to manipulate or deceive people. One of the ways to detect false information is using text processing. Nowadays there are studies directed to the English language to identify deceptive texts, but there are few related works concerning the Portuguese language. In this work, initially, we created a parallel corpus of deceptive book reviews and used some machine learning algorithms to classify deceptive and truthful reviews. A study was made using the research questions proposed by Hauch et al. to do a psycholinguistic analysis of the fake news corpus Fake.Br to verify the most relevant features for fake news classification. Still using the Fake.Br corpus we trained supervised machine learning algorithms to automatically classify fake news and we also use a deep learning algorithm called Hierarchical attention network to verify its performance in fake news detection. (AU) | |
| FAPESP's process: | 18/03129-8 - Analysis and characterisation of deceptive texts written in Portuguese using text processing methods |
| Grantee: | Emerson Yoshiaki Okano |
| Support Opportunities: | Scholarships in Brazil - Master |