Build and publish a graph
Start with RDF data, build a .rete file, then give readers a local file or a
URL. You do not need to run a database service to share it: a suitable static
host can serve the byte ranges that clients request.
Make a file readers can understand
The Quickstart walks through a small build and query. For your own data, include a Dataset Card with its title, licence and source. Readers can inspect the card before choosing what to query.
- Build a .rete, step by step follows one synthetic graph through Python and the CLI, then the card, the schema, the file layout, time and memory.
- Build from Python when the graph comes from your code.
- Import, export and interoperability covers RDF syntaxes, named graphs and exchanging data with other stores.
- Large builds and tabular data explains the choices when the source is a table or the graph exceeds available memory.
Use the CLI reference for exact build flags and defaults.
Put it on a URL
Hosting a graph explains HTTP Range requests and browser CORS settings. Test the published URL with a real query before sharing it; a URL that downloads successfully is not necessarily suitable for browser queries.
Media and SQL companions describes how images, geometry, audio and table exports can accompany a graph. These are optional: start with the graph and add the representations your readers need.
Maintain a published graph
A .rete file is immutable. Rebuilding it or changing its card changes its
identity, even if the data is otherwise unchanged. See
updating a card and
compatibility before replacing a published file.
For applications that need edits between rebuilds, writable graphs explains the manifest and journal approach, including its experimental status and limits.