Diving Deeper: Building Ontologies In The Real World
Ontologies, machine learning, causal inference, and complexity all live in this same neighborhood. Ontologies are the key that opens the door to reliable applications. It is challenging not to dive straight into that, but none of this makes sense when I do.
The introduction section is very high level and will serve as a mental model for several upcoming posts. This post dives into practical applications with a pricing use case right after this first section so stay with me for the introduction. I will quickly move past theory, but the mental model is important.
We could map the universe with an unthinkably complex ontology. This image represents the cosmic web. It is the structure of everything in the universe.
We could map our minds with an equally unthinkable ontology. This image represents the neuron map of a small part of our brain. It is the structure of everything we use to understand the universe.
Nature is repetitive because natural laws enforce a convergence to optimal designs. Anytime we need to express a very complex construct, the graph structure reappears.
Ontologies are naturally built for our minds because they share a common structure. Oddly, it takes a massive amount of learning to translate how our minds see the world into something we can share with others.
When you read the word ‘ontology,’ I want you to see the structure of neurons in our brains. They connect concepts and memories. Connect enough of these, and we form higher-level constructs. Recognizing a familiar face to catching a baseball are all built on this structure.
I also want you to see the cosmic graph. It is easy to forget how much is contained in even the smallest part of the graph when we see it from a high level. Those are strings and clumps of galaxies. The thin sections contain millions of galaxies in any given area. The clumps contain billions of galaxies.
Zoom in on a thin section enough, and we can see a handful of galaxies. Zoom in more, and we see a single galaxy. Further to see stars, a solar system, planet, continent, country, city, neighborhood, crowd, person, and brain. A microscopic sliver of the cosmic web contains another web of extraordinary complexity.
What looks like a simple connection between 2 variables could contain a great depth of additional information. A single variable can be an aggregate of many more or represent a complete ontology. Looking at an ontology can obscure the granularity hiding behind edges and nodes.
The sun and the earth have an ontology-based on gravity. There is another layer explaining the light exchange between them. A single edge connecting the sun and earth in the solar system ontology hides ontologies.
Storing Knowledge In An Ontology
Price is connected to quantity sold. Basic economics states that for most goods, the quantity sold will fall as the price rises. Price elasticity or price sensitivity explains the relationship between a change in price and a change in quantity sold.
Behind every price, there is a product. Behind every unit sold, there is a transaction. Behind the transaction is a customer.
All of this is understood by experts. A person could explain this to me, and I could build an ontology-based on their knowledge. It would look like a database schema or object-oriented programming paradigm.