This study composes of strategic structure for knowledge map generation which is semantically inclined specifically focused on radio-physics as a domain of choice. Radio-physics as a domain of choice has been neglected in the recent time but it is definitely a integral part of radio diagnosis and medical radiology and henceforth there is a need for a knowledge map generation in the web 3.0 using semantically inclined knowledge centric framework. The proposed KMSR model which aggregates terms directly from the dataset and through TF-IDF from the dataset perspective and the generated metadata which is categorized utilizing the Radial Basis Feed Forward Neural Network as a classification of choice the differential partitioning of the classified metadata include three distinct part is quite a novel strategy in the proposed framework and the caption generation using the BERT LLM from the top 10% classes of the top 50% classes that of the classified metadata and the Pearson Correlation Coefficient to classify the dataset using Random Forest light weight machine learning algorithm classification is achieved. Semantic similarity computation through the Wu palmer similarity measure and Resnik similarity measure at various pipeline stages with parts of the obtained classified term from the RBFNN is appreciable and the semantic similarity computation through Wu palmer similarity, Resnik similarity and metaheuristics optimisation using the Emperor Penguin Optimisation algorithm and the knowledge stack based encompassment of entities to formalize the knowledge graph are achieved in the proposed framework with an overall precision of 96.07%, F-measure of 97.19% and the FDR of 0.04 making it the finest in class model for Knowledge Map Generation for Radio-physics as a Strategic Domain.