An automatic model is necessary for creation of ontologies specifically targeting rare domains of international importance like that of island studies. This paper presents a structured framework for generation and synthesis of ontologies for island studies as a domain of choice. It encompasses semantically inclined knowledge driven strategies with a partial learning and reasoning paradigm. The dataset terms are preferentially selected by extracting categories as well as through Term-frequency Inverse-Document Frequency (TF – IDF) model. These terms are used to synthesize the metadata which is further classified using the Generative Adversarial Network (GAN) classifiers. Subsequently, the document dataset is itself classified using the Support Vector Machine (SVM) classifier. Furthermore, semantic relevance is computed by encompassing Lin Similarity measure, Pearson’s Co-relation Co-efficient (PCC) and Normalized Google Distance (NGD) for semantic relevance driven reasoning into the model. Additionally, the process of formalizing an ontology and populating it is accomplished through the utilization of PCC and NGD for semantic reasoning. This model achieves the best accuracy of 96.46% and highest percentage of precision of 95.45% with the lowest value of False Discovery Rate (FDR) of 0.05 making it the best ontology generation framework over all the other baseline models.