Published Online:July 2026
Product Name:The IUP Journal of Knowledge Management
Product Type:Article
Product Code:IJKM020726
DOI:10.71329/IUPJKM/2026.24.3.24-60
Author Name:Spark Tsai
Availability:YES
Subject/Domain:Management
Download Format:PDF
Pages:24-60
Retrieval-augmented generation improves access to enterprise information but typically leaves querytime semantic interpretations dependent on each model invocation. This conceptual study applies artificial intelligence knowledge architecture (AIKA) to enterprise knowledge management through two persistence responsibilities. First, semantic distinctions that have been explicitly established and accepted are externalized as attributes. Second, accepted relationships among semantic units are externalized as persistent relations rather than repeatedly inferred. Claims identify semantic units, while qualifiers, grounding, viewpoints, and derived domains make the resulting knowledge state-bounded, inspectable, and reusable. An enterprise case demonstrates how this architecture converts transient interpretation into governed organizational knowledge across multiple functions. The paper defines a semantic unit as a meaningful assertion available for inference.
Organizations have accumulated large bodies of internal information in requirements repositories, policy documents, contracts, specifications, operating procedures, technical records, databases, meeting records, issue histories, and other enterprise systems. Knowledge management technologies have long attempted to make these materials available for organizational use. More recently, large language models and retrieval-augmented generation (RAG) have substantially changed the way employees interact with such material.