The proliferation of fake news in the digital age has become a critical concern, necessitating effective countermeasures to uphold information integrity and promote media literacy. This thesis addresses the pressing issue of detecting fake news in natural language processing, aiming to contribute to the ongoing battle against misinformation. We delve into the comparative analysis of two prominent methods for fake news detection: the TF-IDF model and Lexical-Based Stance Detection with Logistic Regression.

Our research methodology involves acquiring a diverse dataset from Kaggle, encompassing labeled news articles categorized into various stances such as agree, disagree, discuss, or unrelated. We meticulously preprocess the data, employing techniques like bag-of-words (BoW) and n-gram modeling to extract relevant features while mitigating noise and inconsistencies. Furthermore, we leverage a lexical-based approach to discern the stance of the text regarding specific topics or claims, utilizing sentiment lexicons and dictionaries.

Additionally, we develop and train a sophisticated machine learning classifier to predict text stance based on the extracted lexical features. The performance of our classifier is rigorously evaluated using metrics such as accuracy, precision, recall, and F1-score. By comparing the effectiveness of logistic regression and the TF-IDF model in detecting fake news, we aim to provide valuable insights into the strengths and limitations of these methods for classifying news articles and sentences.

Overall, our research emphasizes the urgent need for reliable fake news detection systems and strives to bridge the gap between information accuracy and reliability in the digital landscape. By evaluating and comparing different approaches, we aim to contribute to the development of robust methodologies that combat misinformation, promote public safety, optimize resources, and restore trust in online platforms. This thesis represents a significant step forward in addressing the challenges posed by fake news and empowering individuals with accurate and trustworthy information.