How This Site Works

This website is organized using a lightweight ontology – a system of typed pages and explicit relationships between them. Every page declares what kind of content it is, and metadata fields encode how pages relate to each other.

This is the same approach I use in my clinical research work, where typed knowledge graphs organize reasoning about patients, treatments, evidence, and assumptions. Here, the same principle organizes a professional life.

The Node Types

Each page on this site has a node-type field in its front matter:

Node Type	What It Represents	Example
person	The site owner	Home, About
project	A body of work	SCD Mortality, TxtArchive
role	A professional position	IU Health, Lucid
publication	A peer-reviewed paper	Listed on Publications
skill	A tool or domain	Cataloged on Skills
writing	An essay or teaching piece	Learning to Change
meta	This page – self-describing	You’re reading it

The Edges (Relationships)

Relationships between pages are encoded as YAML metadata, not as a database:

Relationship	How It’s Encoded	What It Means
uses-skill	`skills: [r, tidymodels]` in a project	This project uses these tools
produced-during	`role: iuhealth` in a project	This project was done during this role
demonstrates	`skills: [statistics]` in a publication	This paper demonstrates this expertise
categorized-as	`categories: [healthcare-ai]`	Standard Quarto cross-reference

An Example

Here’s the front matter for the SCD Mortality project page:

---
title: "SCD Mortality Prediction"
node-type: project
status: active
role: iuhealth
skills: [r, tidymodels, ranger, shap, survival-analysis]
categories: [healthcare-ai, sickle-cell-disease]
---

This single block of metadata tells you:

What it is: a project (not a role, not a publication)
Its status: active (not complete or archived)
Where it happened: during the IU Health role
What tools it uses: R, tidymodels, ranger, SHAP, survival analysis
How to find it: under healthcare-ai and sickle-cell-disease categories

Why Bother?

Three reasons:

Consistency. Every project page has the same structure. Every role page has the same fields. This makes the site easier to maintain and extend.
Discoverability. When Quarto builds the site, listings auto-generate from page metadata. Add a new project page with the right front matter and it appears in the projects listing automatically.
The idea itself is interesting. In my clinical research, I’ve been building ontology schemas to organize reasoning about sickle cell disease – defining what counts as evidence, what counts as a constraint, how assumptions connect to claims. Applying the same principle to a personal website demonstrates that typed knowledge isn’t just an academic exercise. It’s a general-purpose way to organize information.

What This Is Not

This is not a graph database. There’s no JavaScript visualization, no SPARQL endpoint, no runtime query engine. The “graph” exists in the relationships encoded in YAML front matter. Quarto reads these fields and generates the site. The structure is a convention, enforced by consistency rather than software.

This is the same approach as Phase 1 of the SCDCernerProject architecture – convention-based ontology that improves reasoning without requiring infrastructure.

The Schema

The full schema is defined in _ontology-schema.yml at the project root. It specifies which fields are required and optional for each node type, and documents the edge types and their encoding conventions.