Fake News Classifier: By: Pinky Chauhan (University of Illinois at Urbana Champaign)

This objective of this project is to build a classifier system based on machine learning that is able to identify fake news from real/reliable news given a news title and/or news text content as the input. Such a tool can be integrated with social media platforms to flag potentially fake articles or filter those out. This is essentially a data categorization problem where I have trained several classifier models on the following dataset from Kaggle: https://www.kaggle.com/c/fake-news/data After evaluation based on various performance metrics, one of the models (in this case, Linear SVC over unigram bag-of-words/TF-IDF representation) is integrated in the final tester notebook to test with news data. The classifier takes a news article (title and text) as input and provides a prediction for the news article as either of the 2 categories:

• Fake News
• Reliable News

Final Project report/documentation: https://github.com/pinkychauhan89/CourseProject/blob/main/ProjectReport.pdf Presentation:https://github.com/pinkychauhan89/CourseProject/blob/main/Presentation.pptx