ARTIFICIAL INTELLIGENCE FRAMEWORKS FOR DETECTING MISINFORMATION IN DIGITAL INFORMATION SYSTEMS: A LIBRARY SCIENCE PERSPECTIVE ON INFORMATION CREDIBILITY AND TRUST

Abstract

The rapid growth of digital information systems, social media platforms, and online communication networks has accelerated the dissemination of misinformation, creating significant challenges for information credibility assessment, trust management, and reliable knowledge dissemination. Existing misinformation detection approaches frequently rely on isolated content analysis, rule-based filtering, or conventional machine learning techniques that struggle to address the scale, complexity, and evolving nature of modern misinformation campaigns. To address these limitations, this study proposes an Artificial Intelligence (AI) framework for detecting misinformation in digital information systems by integrating advanced natural language processing (NLP), credibility assessment, trust analytics, and Library and Information Science (LIS) principles within a unified detection architecture.

The proposed framework combines semantic text analysis, contextual verification, sentiment analysis, metadata validation, and multi-source credibility assessment to evaluate both content characteristics and source reliability. Core LIS concepts, including authority control, metadata validation, credibility assessment, and trust indexing, are incorporated to enhance information quality evaluation and decision transparency. Experimental validation was conducted using seven heterogeneous benchmark datasets collected from news media, social platforms, academic repositories, and fact-verification sources. The framework employs AI-driven content analysis and source-centric trust evaluation to identify misleading, manipulated, and contextually ambiguous information across diverse digital environments.

Results demonstrate that the proposed framework achieved an overall classification accuracy of 93.7%, outperforming conventional misinformation detection approaches across all major evaluation metrics. Comparative analysis indicates a 31.8% improvement in misinformation identification performance, a 24.6% enhancement in source credibility evaluation, a 21.3% reduction in false-positive classifications, and a 27.9% increase in real-time detection efficiency. Furthermore, the framework supports scalable deployment through a two-tier screening architecture capable of processing approximately 4,200 content items per minute while maintaining robustness against manipulated and contextually ambiguous information. These findings demonstrate that integrating AI-driven analytics with established LIS credibility and trust-management principles substantially improves misinformation detection, information verification, and digital knowledge reliability, providing a scalable solution for digital libraries, social media monitoring, academic information systems, and public information governance.