Optimization and improvement of fake news detection using deep learning approaches for societal benefit
Fake news is a topic that has been discussed for quite some time. Prior to the internet era, it was mostly distributed through yellow journalism, with a focus on sensational news such as crime, rumours, accidents, and amusing news. To rescue the life of people from these fake news propagation, detec...
Guardado en:
| Autores principales: | , |
|---|---|
| Formato: | article |
| Lenguaje: | EN |
| Publicado: |
Elsevier
2021
|
| Materias: | |
| Acceso en línea: | https://doaj.org/article/e42684fa1752492b855e636c44a6d3d5 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| Sumario: | Fake news is a topic that has been discussed for quite some time. Prior to the internet era, it was mostly distributed through yellow journalism, with a focus on sensational news such as crime, rumours, accidents, and amusing news. To rescue the life of people from these fake news propagation, detection of fake news at an early stage becomes the most crucial step. People unknowingly propagate fake news and become a part of fake news propagation. While the original fake news propagators are the one with their aim to target innocent people for spreading the fake news. To stop this series of events, fake news detection and its pattern of propagation becomes very essential to society and the government. Various techniques exist to detect fake news in social media, among which neural networks have shown effective results. For this research, a deep learning based approach has been used to differentiate false news from the original ones. A LSTM neural network has been used to build the proposed model. Besides the neural network, a gloVe word embedding has been used for vector representation of textual words. Also, for feature extraction or vectorization, tokenization technique has been used. N-grams concept is used to enhance the proposed model. The comparative analysis of multiple fake news detection techniques is analysed. The results of proposed model have been evaluated using accuracy metrics. The model outperformed by achieving 99.88% of accuracy. |
|---|