Fighting Against Fake News by Connecting Machine Learning Approaches with Web3

Misleading content, fake news and false media spreading across social media platforms is a threat to society. It negatively effects people and its misuse in political propaganda, cyber crimes and other areas is undeniable. This paper presents how to build a secure, trustful and efficient platform to combat against malicious content and fake news by implementing NLP techniques including stop words removal, topic modelling and by applying machine learning models of KNN, Mulitnomial Naive Bayes and deep learning model of LSTM with Word2Vec and GloVe. These models are fed training and testing data by concatenating two kaggle datasets and selecting sample from them. Their accuracy is also compared at the end. To make the system decentralized Etheruem Blockchain is combined and as an offchain storage for blockchain IPFS is used.