There is a dire need for scientific document recommendation frameworks in the area of Web 3.0. This paper proposes a Web 3.0 compatible semantic-oriented learning how-to reasoning strategy that encompasses ontologies for scientific document recommendation. The proposed model encompasses ontologies using the dataset categories and improving the contextual terms of the dataset in Latent Dirichlet Allocation and ontology alignment using concept similarity with the metadata of the preprocessed query words, metadata generation with each of the aligned ontological entities, and is classified utilizing a powerful deep learning RNN classifier are some of the noble features of the proposed framework. We use the XGBoost classifier, a machine learning feature-controlled classifier to classify the dataset and Lion’s optimization computation using SemantoSim measure and concept similarity helps in standard strategic reasoning through semantic oriented measures and Lion’s optimization serves as a strong metaheuristic optimization paradigm to yield the indicated terms for recommending the scientific documents from the dataset. An overall precision percentage with the recall percentage and the false discovery rate has been attained and, the suggested framework is the best possible model when compared to all other baseline models.