Logistic Regression versus XGBoost: Machine Learning for Counterfeit News Detection

Abstract

In this age of globalization, the unstoppable spreading of fake news via the internet is unstoppable. The spread of false news cannot be supported due to the negative consequences. Society is extremely concerning. In addition, itleads to more serious problems and possible threats, like confusion, misunderstandings, defamation and falsehoods that induce users to share inflammatory content. With the convenience and tremendous increase in information gathering on social networks, it is becoming difficult to differentiate between what is false and what is real. Information can be easily disseminated through sharing, which has contributed to the exponential growth of their forgeries. Machine learning played an important role, in classifying information, although there are some limitations. This article explores various machine learning techniques used to detect fake and fabricated messages. The limitations are discussed using deep learning implementation. In this project, the methodology used is model development and Logistic Regression classifier is considered to detect false news. Based on previous research, this classifier performed well in classification tasks. In this approach, TF-IDF feature is used for the construction of this fake news model to get higher accuracy. The goal of this project is to detect false news using NLP and Machine Learning based on the news content of the article. Following the development of the appropriate Machine Learning model to detect fake/true news, it is deployed into a web interface using Python Flask.