Automated Ontology Construction: Potential for Failure or Success

by
Neuro-Symbolic AI with Ontology

Despite the dramatic advances in LLM and deep learning, modern AI still faces an ‘epistemological crisis’ of hallucination and explainability. This is because the current approach, which relies solely on statistical patterns (correlations) in data, struggles to handle complex business decisions and clear causal reasoning.

This article proposes ‘Ontology’ and ‘Knowledge Graphs’ as solutions, providing an in-depth analysis of how ontology construction—once a labor-intensive failure case—is now being ‘automated’ through integration with the latest LLM technology and Neuro-Symbolic architecture.

It details the evolution toward ‘System 2 Thinking’ AI—capable of logical verification beyond probabilistic guesswork—and ‘Semantic Integration’ that transcends the physical integration limitations of data lakes. It presents concrete technical solutions and future strategies for those seeking to ensure AI reliability and transparency while building truly data-driven intelligent agents.

Decoding Palantir: How “Problem Definition” Bridges Project Management and Ontology

by

A story about the Ontology approach from a PM work perspective, which addresses ‘acceptance criteria management’ and Legacy system integration issues, as well as converting manufacturing floor workers’ tacit knowledge into explicit knowledge and connecting it.

What may sound like a simple consulting statement actually contains deeply embedded, meticulous project management (PM) strategies to prevent large-scale SI project failures, as well as a technical philosophy (Ontology) that has evolved through overcoming complex government and defense data environments.

Today, I would like to reinterpret Palantir’s “problem definition” approach from two perspectives: establishing the PM’s ‘Definition of Done’ and data modeling to overcome legacy environments.

The Great Evangelist Era

by

When you wake up in the morning, a new AI service has emerged into the world, and by lunchtime, YouTube and SNS (Social Network Service) are already plastered with review videos about it.

“This feature is insane,” “You’ll regret it if you don’t learn this now,” “OOO is now obsolete.”

It’s reminiscent of the 15th-17th century ‘Age of Discovery,’ when European powers were desperate to find new continents and plant their flags. I would like to call today’s phenomenon ‘The Great Evangelist Era.’

Evangelist originally means a religious preacher, but in the IT industry, it refers to a ‘technology evangelist.’ Now, not only corporate experts but also countless YouTubers and bloggers have become voluntary evangelists, spreading the gospel of new AI technologies.

I have examined this massive trend in depth from five perspectives: positive, negative, financial, psychological, and knowledge sharing.

Improve Agentic AI Performance Through Understanding Big O

by

Currently, the development of Agentic AI technology is overly focused on implementing functions that can operate in large-scale hardware infrastructure environments. A structural review is necessary to ensure that organizations can guarantee performance at a level that is actually usable when building Agent services internally.
Big O notation is a mathematical notation used to describe the performance of algorithms. Specifically, it predicts the performance of algorithms and architectures in terms of time complexity and space complexity. This review process is essential during system design.