This paper proposes a semantically oriented strategic framework for web video recommendation. The transition of the web from the conventional, standard Web 2.0 to a much more intelligent and semantic Web 3.0 necessitates the utilization of a video recommendation framework that is compliant with Web 3.0. Metadata generation forms the basis of the proposed strategic video recommendation framework, which expands the auxiliary knowledge exponentially within the framework. BERT classifier is integrated into the model to classify the metadata, while Convolutional Neural Network (CNN) is utilised to categorize the dataset. Normalised Pointwise Mutual Information (NPMI) computation serves as a strategic threshold and Genetic Algorithm helps in semantics-oriented reasoning and optimisation through Metaheuristics. Salient Semantic Analysis (SSA) helps in yielding the most optimal final solution set. The proposed framework, the best-in-class framework for video recommendation on Web 3.0, has attained an overall precision of 95.75% with lowest value of FDR of 0.05 and an accuracy of 96.07%.