Abstract:
Advance modern technology and the growth of digitized content (using social media)
manipulation users and turn to fake news, and its impact is deadly. Yellow news coverage
is the cause of another big problem nowadays. Gain popularity and have profited through
clickbait news publisher and social media-based news system circulated unauthorized
news everywhere. The intention is to control religious, political, monetary, and other
genuine things utilizing this simply get to strategy. The biggest concern is that it makes
an annoyance and spreads savagery, even wage wars. Common people are not able to
differentiate between fake and real news. The nature of fake news makes people suspects
genuine news. Advance to use of NLP; it has ended up interesting to look for knowledge
or designs within the era of fake news and thus find better prescient ways to discover fake
news to categorize it from genuine news. In this paper, we propose an ML-based fake
news detection strategy within the Bengali language. The proposed method uses a dataset
on a LR, DT, RF, MNB, KNN, SVM algorithms. The calculation combination of TFIDF-based content features (Unigram, Bigram, Trigram) and vectorizing to include
selection. The accuracy of our proposed model is 92.13 on the Multinomial Naive Bayes
algorithm, which is the highest accuracy from other algorithms. In expansion to this, we
have performed a comprehensive analysis of different machine learning algorithms. At
the same time, we have completed a comprehensive study where we have conducted a
literature review with some questions related to fake news, which has helped us acquire
the knowledge required for this research.
