LEVERAGING EXPLAINABLE AI TO ENHANCE DEEP LEARNING IN FAKE NEWS DETECTION CAPTURING COMPLEX LINGUISTIC PATTERNS AND SEMANTIC INSIGHTS

Abstract

This study looks at how Explainable Artificial Intelligence (XAI) methods can be combined with Deep Learning (DL) models, namely Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN), to make it easier to spot fake news. Fake news is a big problem for society, and regular ways of finding it are not always good at picking up on the complex language patterns and semantic details that are common in false information. Even though DL models, especially CNN and RNN, are very good at classifying text, it is hard to figure out how they make decisions because they are "black boxes." This could make them less reliable in high-stakes situations like finding fake news. To get around this problem, we investigate how XAI methods can make DL models more reliable and useful by making them clear and easy to understand. CNN and RNN models are used in the study to find fake news by looking at the text's complicated language structures and meaning trends. We use XAI techniques like SHAP (SHapley Additive Explanations) and LIME (Local Interpretable Model-agnostic Explanations) to show how these models understand the data they are given, showing important traits and trends that affect the discovery process. Our results show that combining XAI and DL makes the model better at finding fake news by bringing out language clues, rhetorical structures, and semantic errors that are often signs of fake content. The suggested method not only makes detection more accurate, but it also helps us understand the underlying language and semantic patterns better. This makes fake news detection systems more open and reliable. This study tells us a lot about how XAI can make DL models easier to understand. This makes it possible for more useful and trustworthy tools to be used in the fight against fake news.