Sync JSON-LD context with JSON-LD output from vocab
ariutta opened this issue · 6 comments
I had been using a JSON-LD context, but it appears we can generate JSON-LD from our RDFa? Maybe these should be synced.
It would definitely be good to have these synced, otherwise we are prone to them diverging.
There is a recently published working draft on Microdata that has a section on converting between Microdata, RDFa, and JSON-LD.
I started trying to sync the BridgeDb vocab with these three items:
- datasources_headers.txt (see pretty view), which describes datasources.txt (see pretty view)
- Swagger doc for webservice
- JSON-LD context for webservice
Action Items:
- do the
about
anddatatype
columns look correct in datasources_headers.txt? - the response to this webservice query is
ENSG00000160791 Ensembl
. IsENSG00000160791
anidentifier
, anID
or something else? This would be for use in the Swagger doc and JSON-LD context. - what is the IRI for
identifier
(or whatever we want to call it)? Mayberdf:ID
orschema:identifier
(suggested by @AlasdairGray) orbiopax:id
or http://www.rdaregistry.info/Elements/u/#P60052? - for that same query, is
Ensembl
an instance ofbdb:conventionalName
? - if we have multiple instances of CCR5 in a pathway, what name should we use to set
ENSG00000160791
andEnsembl
as properties of each instance? How about{"dataSourceConventionalName": "ENSG00000160791", "dataSourceIdentifier": "ENSG00000160791"}
?
It might be worth using some Schema terms in the JSON-LD, e.g. schema:identifier. You could look at the Bioschemas specifications [fixed link - AR] for suggestions for marking up various types resources.
@AlasdairGray, I support using terms from Schema/Bioschemas.
Action Items
- Do we have a consensus on using terms from Schema/Bioschemas?
- Should there be any
bdb
terms that are exact matches with terms from Schema/Bioschemas? Isbdb:DataSource
an exact match forschema:Dataset
?
bdb:DataSource
and schema:Dataset
are not exactly the same thing. The bdb:DataSource includes notions of the schema:DataCatalog
, which would be the issuer of the identifiers.
The question is then whether they are operationally equivalent for most circumstances?