Bioschemas profile URLs not matched as community-specific standard in R1.3 despite detection in I2

[jeffstazer]

Bioschemas profile URLs not matched as community-specific standard in R1.3 despite detection in I2

Description

Bioschemas Dataset Profile namespaces present in JSON-LD metadata are not being matched as community-specific standards for metric FsF-R1.3-01M-1 ("Community specific metadata standard is detected using namespaces or schemas found in provided metadata"). The standard is detected under I2 (semantic resources) but not under R1.3 (community standards).

Context

The [MIDAS Catalog](https://catalog.midasnetwork.us/) publishes structured JSON-LD metadata for infectious disease modeling resources. Our metadata declares conformance to [Bioschemas Dataset Profile 1.0-RELEASE](https://bioschemas.org/profiles/Dataset/1.0-RELEASE) and includes the Bioschemas namespace in the @context:

"bioschemas": "https://bioschemas.org/profiles/Dataset/1.0-RELEASE"

We also include a conformsTo entry:

{
  "@type": "http://purl.org/dc/terms/Standard",
  "@id": "https://bioschemas.org/profiles/Dataset/1.0-RELEASE",
  "name": "Bioschemas Dataset Profile",
  "description": "A structured data markup specification for describing datasets in the life sciences, extending schema.org Dataset",
  "url": "https://bioschemas.org/profiles/Dataset/1.0-RELEASE",
  "standardType": "metadata schema"
}

Observed behavior

When assessing https://catalog.midasnetwork.us/collection/350 (F-UJI v3.5.1, metric v0.8):

FsF-I2-01M (passes): F-UJI successfully detects the Bioschemas namespace among the active namespaces:

"namespace": "https://bioschemas.org/profiles/Dataset",
"is_namespace_active": true

and

"namespace": "https://bioschemas.org",
"is_namespace_active": true

FsF-R1.3-01M-1 (fails): Despite the above, F-UJI only identifies generic/multidisciplinary standards (Dublin Core, DCAT, Schema.org, PROV). Bioschemas is not matched. The debug log shows:

Found non-disciplinary standard (but RDA listed) -: via ns: Dublin Core - http://purl.org/dc/terms/
Found non-disciplinary standard (but RDA listed) -: via ns: DCAT ...
Found non-disciplinary standard (but RDA listed) -: via ns: Schema.org ...
Found non-disciplinary standard (but RDA listed) -: via ns: PROV ...

No mention of Bioschemas. As a result, only FsF-R1.3-01M-3 (multidisciplinary but RDA-listed) passes at maturity level 1, while FsF-R1.3-01M-1 (community-specific) scores 0.

Suspected cause

The issue appears to be a URL matching gap. F-UJI builds its standards lookup dictionary from FAIRsharing and RDA MSC, normalizing URLs with strip('#/'). Two factors prevent a match:

1. FAIRsharing registers Bioschemas at the project level ([FAIRsharing.20sbr9](https://fairsharing.org/FAIRsharing.20sbr9)), likely with a base URL such as https://bioschemas.org. Our namespace uses the full profile URL https://bioschemas.org/profiles/Dataset/1.0-RELEASE. Since the lookup appears to be exact dictionary key matching rather than prefix matching, these don't match.

2. RDA Metadata Standards Catalog does not include Bioschemas at all (checked at [rdamsc.bath.ac.uk/scheme-index](https://rdamsc.bath.ac.uk/scheme-index)), so there is no fallback path for a match.

Expected behavior

Bioschemas profile URLs (e.g., https://bioschemas.org/profiles/Dataset/1.0-RELEASE) should match the Bioschemas entry in FAIRsharing and be classified as a community-specific (life sciences) standard for R1.3. This could be achieved by:

• Using prefix matching when comparing namespace URLs against the standards registry (so that https://bioschemas.org/profiles/Dataset/1.0-RELEASE matches a registered https://bioschemas.org entry), or
• Registering common Bioschemas profile URLs as additional known URLs for the Bioschemas standard entry

Why this matters

Bioschemas is a recognized community standard endorsed by ELIXIR and the European Research Council. Their own documentation recommends using full profile URLs in dct:conformsTo declarations (see [Bioschemas markup tutorial](https://bioschemas.org/tutorials/howto/howto_add_markup)). Implementers following this best practice will not receive credit for community standard usage in F-UJI assessments.

Reproduction

• URL: https://catalog.midasnetwork.us/collection/350
• F-UJI version: 3.5.1
• Metric version: 0.8
• DataCite support: enabled

Environment

• JSON-LD metadata embedded in HTML via <script type="application/ld+json">
• Full JSON-LD for the test record: https://catalog.midasnetwork.us/jsonld/350.jsonld

[huberrob]

Thanks for the nice JSON-LD example presented in the Dataset!

Indeed bioschemas is not (yet) included in the metadata standards F-UJI recogizes. I'll add this asap. However I am not sure if your example will then be identified as a bioschemas Dataset.

Your JSON-LD clearly indidates that the described resource is of @type schema:Dataset. The record uses exactly one property (conformsTo) which is using a term (value) from bioschemas : https://bioschemas.org/profiles/Dataset/1.0-RELEASE which is the profile you are referring to.
No bioschemas property is used in the JSON-LD. Therefore it would not be classified as bioschemas by F-UJI.

As far as I understood, bioschemas has a dedicated profile for bioschema:Dataset and this should be presented like this:

{
  **"@type": "bioschemas:Dataset",**
  "@context": {
    "schema": "http://schema.org/",
    "rdf": "http://www.w3.org/1999/02/22-rdf-syntax-ns#",
    "rdfs": "http://www.w3.org/2000/01/rdf-schema#",
    "owl": "http://www.w3.org/2002/07/owl/",
    **"bioschemas": "https://discovery.biothings.io/ns/bioschemas/",**
    "bioschemasdrafts": "https://discovery.biothings.io/ns/bioschemasdrafts/",
    "bioschemastypes": "https://discovery.biothings.io/ns/bioschemastypes/",
    "bioschemastypesdrafts": "https://discovery.biothings.io/ns/bioschemastypesdrafts/",
    "bioschemasdeprecated": "https://discovery.biothings.io/ns/bioschemasdeprecated/",
    "dct": "http://purl.org/dc/terms/",
    "@dateModified": "01/19/2024, 16:03:00",
    "dwc": "http://rs.tdwg.org/dwc/terms/"
  },
  "name": "Test",
  "description": "tests"
}

At least the Bioschemas generator https://discovery.biothings.io/markup-generator?schema_class=bioschemas:Dataset creates this.

In summary: the bioschemas namespaces needs to be listed in the @context definition and at least one bioschemas property needs to be used . Otherwise F-UJI and other parsers will only be able to identify/parse properties which comply with the schemas indicated in @context

[jeffstazer]

Thank you for the quick fix on adding Bioschemas to the standards list, and for the clear explanation of the detection mechanism.

I understand the challenge — the Bioschemas Dataset Profile (1.0-RELEASE) defines which Schema.org properties to use for life sciences datasets rather than introducing its own namespace properties, so namespace-based detection won't find it. Their markup tutorial recommends @type: "Dataset" (Schema.org) with dct:conformsTo pointing to the profile URL, which is the pattern we followed.

I've raised this with the Bioschemas community at BioSchemas/specifications#718 to get their input on how tools like F-UJI should detect Bioschemas conformance.

[huberrob]

@jeffstazer I just have seen that using dc:conformsTo is also used by Croissant to indicate such schema.org variants so I tend to include this property as additional source of types similar to schema:additionalType despite the different semantics