Global Journal of Computer Science and Technology, C: Software & Data Engineering, Volume 22 Issue 2

Fake News Detection: Covid-19 Perspective Global Journal of Computer Science and Technology Volume XXII Issue II Version I 8 Year 2022 ( ) C © 2022 Global Journals three sets. But the poor value in recall (64%) and f1- score (76%). But in set 2 and set 3 it gives better results on all sections. Passive aggressive classifier using count vectorizer on Table 3. In this part set 2 and set 4 both shown the highest accuracy of 90% then other two sets. But set 4 has a poor outcome on recall (68%) and f1- score (73%) while detecting fake news than other sets, that we have tested. The other three set shown decent outcomes on all three parameters. Logistic regression also given the highest accuracy of 91% in set 4. But not a satisfying outcome on the other three-parameter in detecting fake news. In general, set 2 given the best performance out of other four sets, having an accuracy of 89%. In set-4, Fake news detection Fig. 10: Fake News Detection Comparison Between Four Algorithm Using Set-2 Multinomial Na¨ıve Bayes Passive Aggressive Classifier Logistic Regression Support Vector Machine 0 20 40 60 80 Percentage Accuracy Precision Recall F1-score (a) Performance Analysis: Fake News Detection using Countvectorizer Multinomial Na¨ıve Bayes Passive Aggressive Classifier Logistic Regression Support Vector Machine 0 20 40 60 80 100 Percentage Accuracy Precision Recall F1-score (b) Performance Analysis: Fake News Detection using TF-IDF