A Hybrid Metadata-Intelligent Framework for Fake News Detection, Ranking, and Web Preservation

The explosive growth and ephemeral nature of digital news content pose significant challenges to its long-term preservation and credibility verification. To address this, we propose an Enhanced Web Preservation Framework (EWPF) that integrates credibility assessment as a core component of the digital archiving process. The research unfolds in five distinct phases. First, a Multivocal Literature Review (MLR) was conducted to extract and validate 14 critical credibility factors (CFs) such as source reliability, author reputation, social media footprint, and topical relevance. In the second phase, these CFs were embedded into the EWPF through a Credibility Ranking Algorithm, using the Analytical Hierarchy Process (AHP) for factor weighting. A custom dataset comprising 800 annotated news articles was constructed in the third phase, tailored to represent real-world variations in digital news credibility. In the fourth phase, multiple Machine Learning (ML) models were applied to classify news content based on credibility, achieving competitive accuracy and low error rates. An ablation study was performed to systematically evaluate the individual contribution of each credibility factor, revealing key factors with significant impacts on overall framework accuracy. Finally, an interactive web interface was developed and integrated with the EWPF to visualize dynamic credibility scores and facilitate user access to preserved, high-quality news content. Experimental evaluations demonstrate that the EWPF enhances the accuracy of credibility-aware archiving while maintaining computational efficiency and minimal storage overhead. This research presents a novel, end-to-end framework that bridges credibility analytics with long-term web preservation, offering a scalable solution for combating misinformation in digital journalism.