AI initiatives often face a tradeoff between accuracy, transparency, and governance. While knowledge models and taxonomies provide trusted business context and explainability, modern AI offers powerful capabilities for classification, matching, and discovery. This session explores how these approaches can be combined to deliver more trustworthy and effective AI systems.
The Progress Data Platform AI Data team will present exploration in three areas: Concordance – An evaluation of entity relationship generation using classical OWL/SKOS knowledge models and Semaphore-style extraction techniques. We will share benchmark results, findings, and lessons learned on creating reliable semantic foundations for AI. Classification – A cross-validation approach that combines client data and OWL/SKOS taxonomies with embeddings and Small Language Models (SLMs) to evaluate how structured knowledge and AI can improve classification quality and confidence. Matching – An exploration of hybrid matching techniques that combine transparent lexical matching with contextual embeddings. By reinforcing semantic search with deterministic validation, we assess how missing concept and entity relationships can be identified while maintaining traceability and trust.
Attendees will gain insight into emerging approaches that blend symbolic AI and modern AI to improve the accuracy, and governance of future enterprise AI solutions.
