Development of a system to analyze fake news

Abstract

False information has the power to mislead people, change public opinion, and erode trust in the media. Fake news can have major implications, including negative emotional, financial, and economic effects. These impacts underline the need of addressing and mitigating the dissemination of inaccurate information, as it has an impact on people, markets, and society.

The objective of this project is to develop a web application integrated with machine learning and deep learning models to detect fake news articles. Numerous machine learning models and neural networks were explored. The models were also tested with news articles gathered from news sources like The Straits Times, BBC, CNN, Channel News Asia (CNA) and CNBC. The models were chosen by conducting a thorough analysis of multiple data sources and extensive research.

To obtain a deep understanding of the features of the data, a thorough investigation of the chosen ISOT fake news detection dataset was carried out using Exploratory Data Analysis (EDA) techniques.

The analysis of the fake news articles in the web application has also expanded with the inclusion of sentiment analysis, text summarization and Named Entity Recognition. The inclusion of sentiment analysis would help the user discern whether the article conveys positivity, negativity, or neutrality, providing insight into the potential intent of the article.