Requirements
IRI minting
- ideally there could be some kind of signature that affirms "the dataset published to this IRI was authorized by a person who has control of the namespace" but I guess its existance at that URL speaks for itself.
- see 2024 10 25 Semantic Flow Overview for approaches to trust
Dereference Pages
- for Reference Pages you don't want all assertions, only some of the most useful ones, perhaps with links to where you can get more (ie., catalog links)
- mapping of templates to facets, namespaces, datasets, sparql queries
Canonicalization
- to make diffing easier, all datasets should be canonicalized before commit
System Time
There are multiple levels of "this data was added, changed, or tombstoned":
Relator Atoms
Named Individuals
- the act of using/declaring an IRI itself is a data point. It generates something into existence, even if it doesn't play a role in a relator. And possibly we'll want to be able to issue statements with that IRI that don't use relators.
- optionally, you can use
Version-On-Commit
Backlinks