In today's digital landscape, a specialized recommendation system addressing specific fields like geology and geography is urgently required. These domains have been underserved in terms of contemporary Web 3.0 standards, which prioritize semantic understanding and knowledge-centric recommendations. To bridge this gap, a tailored recommendation system has been introduced for geography and geology. This system employs a strategic approach to progressively enhance its knowledge by categorizing dataset information. It achieves this by enriching data with Linked Open Data Cloud and WikiData, followed by meta tag creation and deep learning transformer-based content classification. This transformation aims to improve content comprehensibility through a three-step process involving Google Transcript API, Nell, and Linked Open Data Cloud. The system also calculates concept convergence using explicit semantic analysis and selects pertinent features via the Petraitis index. These features are employed by a Bagging classifier, which utilizes KL divergence, Itakura-Saito distance, and the Akaike information criterion to identify related terms. This leads to the generation of more relevant facets for web page recommendations. Impressively, the proposed system achieves high precision and f-measure scores, coupled with a notably low false discovery rate. The paper presents an evaluation conducted on standard integrated datasets, achieving an overall highest average precision of 95.59%, highest average recall of 97.09%, highest average accuracy of 96.34%, highest average F-Measure of 96.3341613% and with the lowest value of FDR of 0.05.