Every page here, and every link between them
The whole site as one figure: each page a mark, each internal link a line, grouped by the section the page belongs to. The same information follows as a list, so nothing here depends on reading a picture.
One dataset, two forms
The figure is the shape of the site. The list below it is the same dataset in the form an agent, a screen reader or a small screen can use directly: every page, the section it belongs to, and how many pages link to it. Nothing in the figure is missing from the list, and both are generated from the same pass over the site's files.
This page is the site's own index of itself. The list below is plain HTML with real links and no script; the figure above is inline SVG, so its labels are text. Machine-readable indexes: agents.md · llms.txt · sitemap.xml.
The site as a graph
Every page on the site is a mark and every internal link between pages is a line. Each cluster is one section, named in its own colour.
Every page, by section
The same data as the figure, in text, and this page included. A count says how many other pages link to that page from inside their own body; links in the shared header and footer are not counted.
Knowledge
7 pages, 21 links between them.
- Ontologies — 14 pages link here.
- Taxonomies — 13 pages link here.
- Knowledge formats — 12 pages link here.
- Knowledge graphs — 11 pages link here.
- Knowledge — 7 pages link here.
- Knowledge bundles — 5 pages link here.
- DIKW as an audit ladder — 4 pages link here.
Research
5 pages, 13 links between them.
- Research — 26 pages link here.
- Agentic AI — 6 pages link here.
- Semantic Communication — 5 pages link here.
- Memory for Assistants and Agents — 4 pages link here.
- World models — 4 pages link here.
Standards
24 pages, 96 links between them.
- Standards — 46 pages link here.
- W3C SKOS — 12 pages link here.
- RDF — 11 pages link here.
- PROV-O — 9 pages link here.
- JSON-LD — 8 pages link here.
- OWL 2 — 8 pages link here.
- SPARQL — 8 pages link here.
- Memento — 7 pages link here.
- Turtle — 7 pages link here.
- RDFS — 6 pages link here.
- Apache Parquet — 5 pages link here.
- ISO 25964 — 5 pages link here.
- llms.txt — 5 pages link here.
- Apache Iceberg — 4 pages link here.
- GraphQL-LD — 4 pages link here.
- HDT — 4 pages link here.
- Palantir Foundry Ontology — 4 pages link here.
- ServiceNow CSDM and CMDB — 4 pages link here.
- Snowflake — 4 pages link here.
- Databricks — 3 pages link here.
- Delta Lake — 3 pages link here.
- Microsoft Graph — 3 pages link here.
- Neo4j — 3 pages link here.
- Salesforce Platform and Data Cloud — 3 pages link here.
Agents
3 pages, 6 links between them.
Adoption
5 pages, 14 links between them.
- Adoption — 19 pages link here.
- Companies ahead in the game — 4 pages link here.
- Knowledge engineering by domain — 4 pages link here.
- Library and information science — 4 pages link here.
- Enterprise business, general — 3 pages link here.
Learn
4 pages, 9 links between them.
- Learn — 4 pages link here.
- Course terms — 3 pages link here.
- Courses — 3 pages link here.
- Glossary — 3 pages link here.
Other
20 pages, 85 links between them.
- Context Engineering — 14 pages link here.
- Why Knowledge — 14 pages link here.
- Blogs — 12 pages link here.
- Enterprise — 12 pages link here.
- Contact — 9 pages link here.
- Token Savings & Grounding — 8 pages link here.
- Free resources — 7 pages link here.
- Personal Assistant — 5 pages link here.
- Training — 5 pages link here.
- Internship — 4 pages link here.
- About — 3 pages link here.
- Context is not a bucket — 3 pages link here.
- Doctoral research — 3 pages link here.
- Grounding is the flip side of token savings — 3 pages link here.
- What enterprise agents actually need from knowledge — 3 pages link here.
- Why knowledge fails first in agentic systems — 3 pages link here.
- How this site is made — 2 pages link here.
- Community — 1 page links here.
- Knowledge Sidekick — 1 page links here.
- Site map — 1 page links here. This is the page you are reading, and it is not drawn in the figure above.
Home
1 page, 0 links between them.
- Knowledge Sidekick — 33 pages link here.
How this figure is drawn
Nothing here is placed by hand. A script reads every page in the site's files, records the links inside the page body, and lays the result out; the clusters, the list and every count on this page come from that one pass, so the figure cannot disagree with the list under it.
Links in the shared header and footer are excluded. They appear on every page, so counting them would make every page look equally connected and hide the structure that is actually there.
Position encodes section, not link density. Each section is placed as its own region so the clusters stay separable; a layout driven by link strength puts the sections on top of one another and reads worse at this size. The link counts are carried by the mark size and by the list instead.
The figure is inline SVG, not an image, so every label is real text and survives a crawler, a screen reader and print. No diagram library is loaded: the site's only external script is the navigation, and the zoom on this page is a few lines inline.
The marks are small, and WCAG 2.5.8 asks for a target of at least 24px or sufficient spacing around it. The list above is the equivalent control: every page in the figure has a full-size link there.
FAQ
Every page on this site and every internal link between pages, grouped by the section each page belongs to. The list underneath carries the same information, with a link and an inbound count for every page.
Because the first question a map answers is where a page sits. A layout driven by link strength alone interleaves the sections and the grouping stops being readable, so each section is placed as its own region. The trade is that position shows section, not closeness.
No. The whole figure is inline SVG and renders with no script at all; the script on this page only adds zoom and pan. Every page in the figure is also a plain link in the list below, so the information survives if the figure is not readable at all.
Marks that carry more links in this figure are labelled at full strength and the rest are dimmed, rather than removed. The names stay in the file, so a crawler, a screen reader and a print engine still get every label.
Page updated .