A strategic model for image tagging based on metadata augmented ontology derived from Wikidata has been put forth as there is a need for Web 3.0 Compliant Frameworks as Web 2.0 is transfiguring into a Web 3.0.The proposed framework strategically encompasses Semantic Intelligent scheme with its hybridization with Learning Paradigms like the AdaBoost Classifier to classify the dataset and Recurrent Neural Networks to classify the Metadata generated. Synonimization with Wikidata Augmented Entity enrichment facilitates in yielding incremental knowledge to derive a context tree and Generate intermediate Ontologies. The framework also encompasses DBScan Clustering using Second Order Co-Occurrence Pointwise Mutual Information The framework uses the Kullback divergence with the step variance of 0.10 to select the features effectively and then classifies the image datasets using the support vector machine and the generated metadata data using recurrent neural net. The normalized pointwise mutual information measures with a median threshold of 0.50 are computed to assess semantic relationship between classified instances and an initial set of solutions is obtained which is then optimised using ant colony optimisation, where Horns index is used as the objective and a step deviance of 0.15 is used. The proposed OWITD model has been provided with a large amount of experimental validation to show that it is able to achieve impressive performance rates of 96.81. The knowledge-based method with its fusion in ontological reasoning, combination of classification schemes and nature-based maximization, therefore, shows high effectiveness in the case of semantic-web image-tagging.