Title: Artificial Intelligence Across Disciplines: Research, Innovation, and Intelligent Solutions
Editors: Dr. Subita Bhagat, Dr. A. Balamurugan, Dr. P. Krishna Kumar, and Mrs. S. Nandhini Devi
ISBN: 978-81-69857-83-3
Chapter: 24
DOI: https://doi.org/10.59646/815/24
Authors: Dr. A.C. Santha Sheela, and Dr. V. Ragavi
Abstract
This chapter examines artificial intelligence (AI)-powered semantic computing and machine knowledge representation, focusing on how hybrid neural-symbolic architectures resolve representation bottlenecks across enterprise domain graphs. While classical ontologies ensure strict logical consistency, they suffer from brittleness and computational intractability when scaled to large, noisy datasets. Conversely, statistical continuous-space embeddings capture latent relational topology efficiently but operate as opaque black boxes prone to ungrounded semantic drift. To bridge this dichotomy, this chapter establishes a neuro-symbolic framework coupling Description Logic axioms with complex vector space embeddings. Using biomedical multi-relational graphs as a rigorous case study, we demonstrate formal representation translation, non-Euclidean manifold learning, and rule-guided link prediction. Empirical validation reveals superior inferential precision, higher mean reciprocal ranks, and resilient link completion over noisy entity spaces. The chapter finishes by formalizing end-to-end evaluation metrics, computational algorithms, and industrial deployment paradigms.
Keywords: Semantic Computing, Machine Knowledge Representation, Knowledge Graph Embeddings, Neuro-Symbolic Artificial Intelligence, Relational Link Prediction, Description Logics.