Structure
Knowledge graph
Every paper, idea and note declares what it relates to; that is the only structure the library keeps, and the graph is what those declarations look like when drawn.
Edges come from three places: an explicit relation in frontmatter, a citation of a reference, and the topic or tag a document is filed under. Nothing is inferred and nothing is stored twice — change a paper’s frontmatter and the graph follows on the next build.
- Topic
- 9
- Paper
- 2
- Idea
- 0
- Research note
- 0
- Reference
- 33
- Tag
- 10
- Relations
- 19
Most connected
The nodes with most relations attached, which is usually a good place to start reading.
Most connected papers
Busiest topics
- Architecture2 links
- AI1 link
- Cloud1 link
- DevSecOps1 link
What this is not
The graph is a view of the library, not a search engine over it. It does not rank relevance, it does not learn, and it holds no state: reload the page and the layout is recomputed from the same declarations.
What it does replace is a folder tree. A paper can belong to several topics, cite several references and be picked up by several ideas at once, which a hierarchy cannot express without duplicating files.