Linking Scattered Knowledge Together

The idea

Most organizations keep knowledge in separate places: a schedule here, a parts list there, test results somewhere else. Each record makes sense alone, but the connections between them live in people's heads. A knowledge graph writes those connections down. Things become nodes, relationships become named links (Networks: The Math of Connections), and questions become walks along the links.

Why it attracts me

I want tools that explain, not just retrieve. A search returns documents. A graph can answer which later jobs are held up by a late part, because it knows what depends on what. It must also keep track of where each fact came from. A graph that merges many sources without that record quickly becomes hard to trust.

An example

This map is itself a small knowledge graph. Each idea is a node, each link has a type and a written reason, and each idea is labeled for the kind of claim it makes. The labels and reasons are there so that a link between systems and a shared consciousness (Many Minds, One Shared Awareness) shows up as a loose picture, not as evidence.

Where it connects

At a larger scale, I see knowledge graphs as practical attempts to bring together understanding that is spread across many people. That links them, loosely, to a far more speculative image I hold of a universe learning about itself (A Universe Learning About Itself). Grounded data about real supply links also matters for long-term investing (Investing in Long-Term Industry Shifts).

Questions I'm still exploring

• How do I keep track of where every link came from as sources are merged?
• Which important relationships never get written down anywhere?
• Can a map of my own ideas, like this one, stay honest as my thinking changes?

Sources and further reading

• Aidan Hogan et al., "Knowledge Graphs," ACM Computing Surveys 54(4), 2021 — Open-access arXiv version.