There exists a growing need for large-scale ontology synthesis across specialized domains of socioeconomic and societal importance within the Web 3.0 environment. This paper presents a strategic framework for ontology synthesis dedicated to Ambedkar and Peace Studies as a domain of choice. The proposed system operates on the principle of extracting terms and categories from domain-specific datasets to establish a foundational semantic structure. Metadata generation is achieved using OpenKLAS, followed by classification through DistilBERT, a deep learning model capable of automatic feature extraction and contextual classification. The framework then advances toward seed knowledge generation using Latent Semantic Indexing (LSI), integrating the enriched outputs from large language models such as Llama-4-Scout and Mistral-Medium-3 to enable generative knowledge expansion. At multiple stages of processing, semantic reasoning is quantified using the Grubel–Lloyd Index and Adaptive Pointwise Mutual Information (APMI), both enhanced with Jackknife resampling to reduce bias and maintain diversity in entity selection. The optimization of ontology alignment is performed using the Coyote Optimization Algorithm, ensuring an optimal balance between internal semantic coherence and external conceptual diversity. A lightweight AdaBoost classifier further refines the categorized entities to enhance accuracy. The resulting ontology demonstrates improved structural consistency, semantic integrity, and reasoning efficiency, outperforming existing baseline models. The framework achieves an average precision of 96.98%, recall of 97.64%, accuracy of 97.31%, and a false discovery rate (FDR) of 0.04, confirming its effectiveness in synthesizing domain-specific ontologies for Ambedkar and Peace Studies.