In response to the burgeoning data sets dedicated to space science within the web 3.0 framework, this paper addresses the need for a strategically devised semi-automatic ontology generation model. This intricate framework focuses on extracting data from datasets dedicated to space science, meticulously classifying it using a deep learning transformer-based classifier and utilizing the top 10% of classified instances as a federated feature for another auxiliary classifier, XGBoost. The model seamlessly integrates knowledge resources beyond metadata, drawing from semantic wikis like Google and Wikidata. Employing the encoding similarity index with empirically determined thresholds and the Flower Pollination Algorithm as the selected metaheuristics technique, the model generates optimal solution terms. Achieving an impressive overall performance characterized by a notable recall and low False Discovery Rate (FDR), this work establishes the model as a best-in-class solution for ontology generation, particularly within the space science domain in the web 3.0 context.