NAOMS · building block
Knowledge graph
Everything a person has becomes one graph you can search by meaning.
The knowledge graph is the one graph every content type, notes, photos, contacts and more, reads and writes into, implemented in the graph and graph-browser packages. It supports exact-match text search alongside meaning-based search over embeddings on the same data, and matching content is also fed into a standards-based triple store so it is available as linked, JSON-LD-shaped facts, not only through the graph's own query API.
What a person notices
- Searching for “the trip we planned last spring” can turn up the right notes, photos and messages even when none of them use those exact words.
- Notes, photos, contacts and everything else sit in one place, instead of a separate silo per app with its own search box.
- Someone else’s app, built by a different developer, can surface something the person already saved elsewhere, because it’s the same graph underneath.
What you get as a developer
The knowledge graph is the one place content of every kind ends up, queryable by exact match, by meaning, or as linked standards-based data.
- Notes, photos, contacts and other content types all read and write the same underlying graph, so a query doesn’t need to know which feature originally created a given piece of data.
- Search combines exact-match text search with meaning-based search over embeddings on the same data, so a query can be a keyword or a rough description and still find the right thing.
- Alongside the main graph, matching content is also fed into a standards-based triple store, so data is available as linked, JSON-LD-shaped facts for tools that expect that shape, not only through the graph’s own query API.
- How a node gets turned into searchable text is defined per content kind, so a new package can teach the search engine how to read its own data without touching the search engine itself.