News classification and fake news detection based deep learning and NLP techniques

• Main Author: Dee, Yi Cheng
• Other Authors: Mao Kezhi
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science; Fake news detection; News classification; Deep learning; NLP techniques
• Online Access: https://hdl.handle.net/10356/176355
• Institution: Nanyang Technological University

As the saying goes information is power and this cannot be truer especially in the digital age of the 21st century we are living in. Vast amounts of information in the form of news and posts are flowing into the internet at a tremendous rate every second worldwide. However, the reliability of the news cannot be guaranteed as there are those who seek to manipulate the masses for achieving their own or organization selfish goals. Thus, news classification and fake news detection are vital pillars in managing the deluge of information in this digital era. The main objective is to develop a Python based tool based on Natural Language Processing (NLP) and Deep Learning techniques to classify and authenticate news articles. This can be simplified down to a classification problem using multiple models for more conclusive results. To achieve this objective many powerful python libraries have been utilized. Some of the libraries that have been used for result visualizations are Numpy, Seaborn, and Pandas. For Deep Learning more advanced libraries have been utilized such as Keras, TensorFlow, and Pytorch. This Final Year Project (FYP) aims to identify the best machine learning model in classifying and detecting fake news through extensive research, analysis, and experimentation with hopes of being able to identify and reduce the spread of fake news before any undesirable consequences occur. The author of this report has conducted thorough research and review of multiple data sources and applied various exploratory data analysis techniques to filter out biased datasets and information. This report delves into the pre-processing steps of the dataset, building, training, fine tuning of various models and interpretation of results. This FYP enabled the author to gain deeper and more comprehensive insights into many different pre-processing techniques as well as multiple machine learning models and out of which combinations will yield the highest accuracy for news classification and fake news detection. Overall, this FYP serves as an important step towards raising awareness about fake news as well as to detect and reduce the negative impact of fake news in the world.