The rise of fake news is one of the downsides of the internet age and concern over the problem is global. Aggravating the problem, much information remains unknown regarding the vulnerabilities of individuals, institutions and society to manipulations by malicious actors. Resolving the problem using natural language processing is a hot topic of discussion among data scientists.
In this project, data consists of claims, articles and its metadata. Each claim is a sample of data set.
Three approaches have been adopted:
- Passive aggressive classifiers
- LSTM
- BERT using TPU
Link to data: https://drive.google.com/open?id=1eEBHNk9nyi7W4lydsZGZWP2SrebRB9hR