An Ensemble Deep Learning Technique to Detect COVID-19 Misleading Information
| Autor: | Fayez Gebali, Kin Fun Li, Mohamed K. Elhadad |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
Word embedding
Recall Computer science business.industry Deep learning media_common.quotation_subject Word error rate 02 engineering and technology Machine learning computer.software_genre 030218 nuclear medicine & medical imaging 03 medical and health sciences 0302 clinical medicine 0202 electrical engineering electronic engineering information engineering Preprocessor 020201 artificial intelligence & image processing Quality (business) Misinformation Artificial intelligence business computer Word (computer architecture) media_common |
| Zdroj: | Advances in Intelligent Systems and Computing Advances in Networked-Based Information Systems Advances in Intelligent Systems and Computing ISBN: 9783030578107 NBiS |
| ISSN: | 2194-5365 2194-5357 |
| DOI: | 10.1007/978-3-030-57811-4_16 |
| Popis: | This paper aims to combat the Infodemic related to COVID-19. We propose an ensemble deep learning system for detecting misleading information related to COVID-19. This system depends on the shared COVID-19-related information from the official websites and Twitter accounts of the WHO, UNICEF, and UN, as well as the COVID-19 pre-checked facts from different fact-checking websites, as a source of reliable information to train the detection model. We use these collected data to build an ensemble system that uses several deep learning techniques to detect misleading information. To improve the performance of the proposed ensemble detection system, we implement a data preparation and preprocessing step, along with a features engineering step. We deploy Word Embedding based on a pre-trained word embedding list in addition to the existing word impeding in the input layer of the employed techniques. The experimental results are examined using fourteen performance measures (Accuracy, Error Rate, Loss, Precision, Recall, F1-Score, Area Under the Curve, Geometric-Mean, Specificity, Miss Rate, Fall-Out Rate, False-Discovery Rate, False-Omission Rate, and the Total Training Time). The obtained results are promising and indicate the quality and validity of the trusted information collected, for building misleading-information detection systems. It is worth noting that, in this paper, we use the terms “misleading information”, “misinformation”, and “fake news” interchangeably. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|