Explainable tweet credibility ranker: A comprehensive credibility solution

It is observed that previous credibility assessment studies missed some important concepts of credibility. Their data was self-extracted and annotated only on the basis of basic human perception. Their decision was not explainable and led to dissatisfaction. The study attempted to accomplish all these objectives through an explainable and comprehensive credibility solution. The solution adheres to all necessary concepts of credibility presented in the microblog credibility framework and ranks the tweet from 0 to 5. The decision is also explainable to the user. The standard dataset is used with further credibility ranking, which is done by the experts using guided annotation. Despite the known trade-off that exists between accuracy and explainability, we have used all those models which produce high accuracy but our solution is also explainable. Light GBM performed the best. Results are also compared with baselines, which demonstrate that our method produced good results with an Acc. of 99.4.