Abstract
Recently, fake news has become a serious problem in our society majorly
due to the cheap and easy availability of social media at every corner of the world. The
widespread dissemination of false news has the potential to have a variety of harmful
consequences for people and society. Hence, many researchers are finding different
ways to detect fake news in a given news corpus. So here we came up with the idea of
fake news detection using machine learning that detects fake news over the real news.
During pandemic, fake news detection played an important role. Detection and
identification of fake news in the social media, or any related news channels has played
a major responsible sector to avoid unnecessary panic situation in mankind.
This paper is aimed at developing a Machine Learning model for deception detection
using Natural Language processing techniques and machine learning algorithms. It
detects fake news that comes from non-reputable sources which mislead people and
distracts them with various fraud messages and unnecessary texts, by building a model
using count vectorise, TF-IDF and logistic regression algorithm. Using this algorithm,
the proposed technique identifies and rectifies real and fake news and this is an
important sector during the pandemic situation.
However, there is difficulty in choosing the right metric for the evaluation of the
model. Classification accuracy is one of the most used metrics to detect the
performance of the model, in this paper we consider the parameters such as F1 score,
confusion matrix, precision and recall. Abstract environment.