Possible bug converting to RDF

[nein09]

Hi. I've noticed a bug in a tool that uses json-gold (gleanerio/gleaner#79) and I'm wondering if this is a json-gold bug or just me not using json-gold correctly.

What happens is that I have some json-ld that I want to turn to RDF, let's say this:

    "@context":
    {
        "@vocab": "https://schema.org/"
    },
    "@type": "Dataset",
    "@id": "ad780d9a-e6a7-4a6f-824c-367b40d377b8",
    "name": "ACAP Saint John: Temperature Logger Data",
    "description": "ACAP Saint John’s Water Quality Monitoring Programs aim to build upon the community-based water monitoring program established in 1992 in the Greater Saint John Area. The data collected from these various monitoring programs is used to track stressors, determine restoration sites, strengthen/shape policy, and expand community partnerships. Water temperature is logged in various watercourses across the Greater Saint John using temperature logger deployed for months at a time.",
    "url": "https://datastream.org/dataset/ad780d9a-e6a7-4a6f-824c-367b40d377b8",
    "version": "2.0.0",
    "datePublished": "2019-02-04T22:19:58.129Z",
    "dateModified": "2022-04-30T15:48:42.840Z",
    "isAccessibleForFree": true,
    "keywords": "Harbour, Saint John, New Brunswick, Water temperature, Urban, Bay of Fundy",
    "license": "https://opendatacommons.org/licenses/by/1-0/",
    "citation": "ACAP Saint John. 2022-04-30. \"ACAP Saint John: Temperature Logger Data\" (dataset). 2.0.0. DataStream. https://doi.org/10.25976/12cf-1g82.",
    "identifier":
    {
        "@type":
        [
            "PropertyValue",
            "datacite:ResourceIdentifier"
        ],
        "datacite:usesIdentifierSchema":
        {
            "@id": "datacite:doi"
        },
        "propertyID": "DOI",
        "url": "https://doi.org/10.25976/12cf-1g82",
        "value": "10.25976/12cf-1g82"
    },
    "temporalCoverage": "2016-08-04T13:54:00-03:00/2018-12-06T10:55:50-04:00",
    "spatialCoverage":
    {
        "@type": "Place",
        "geo":
        {
            "@type": "GeoShape",
            "box": "-66.19002 45.193191 -65.88223 45.40079"
        }
    },
    "measurementTechnique": "Water temperature is logged using either HOBO or EasyLog temperature loggers attached to cinder blocks and left in the watercourse for various time periods based on the program.",
    "variableMeasured":
    [],
    "creator":
    {
        "@type": "Organization",
        "name": "ACAP Saint John"
    },
    "publisher":
    {
        "@type": "Organization",
        "name": "DataStream",
        "url": "https://datastream.org",
        "logo": "https://datastream.org/favicon.svg"
    }
}

The code that processes it is a bit complicated but I'll summarize what happens with it here:

	var myInterface interface{}
	err := json.Unmarshal([]byte(jsonld), &myInterface)

	proc := ld.NewJsonLdProcessor()
	options := ld.NewJsonLdOptions("")
        options.Format = "application/nquads"

        nq, err := proc.ToRDF(myInterface, options)

	return nq.(string), err

And what I'm getting back out is an incomplete RDF representation of the data. In this instance, I get

_:b0 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://schema.org/Organization> .
_:b0 <https://schema.org/name> "ACAP Saint John" .
_:b1 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://schema.org/PropertyValue> .
_:b1 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <datacite:ResourceIdentifier> .
_:b1 <datacite:usesIdentifierSchema> <datacite:doi> .
_:b1 <https://schema.org/propertyID> "DOI" .
_:b1 <https://schema.org/url> "https://doi.org/10.25976/12cf-1g82" .
_:b1 <https://schema.org/value> "10.25976/12cf-1g82" .
_:b2 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://schema.org/Organization> .
_:b2 <https://schema.org/logo> "https://datastream.org/favicon.svg" .
_:b2 <https://schema.org/name> "DataStream" .
_:b2 <https://schema.org/url> "https://datastream.org" .
_:b3 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://schema.org/Place> .
_:b3 <https://schema.org/geo> _:b4 .
_:b4 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <https://schema.org/GeoShape> .
_:b4 <https://schema.org/box> "-66.19002 45.193191 -65.88223 45.40079" .

You can see that lots of top-level properties, like the type DataSet, name and description are missing.

Is this a bug in json-gold, or something I'm doing wrong? Do you have any tips or tricks I could try?

Thank you.

[kazarena]

@nein09, my initial observation is that your context doesn't define all the required properties that are present in the document, and hence these properties get dropped at JSON-LD expansion stage. I would recommend debugging it by invoking the Expand() function first and analysing the output. You may also use Safe Mode (see here). It will prompt you about any unexpected properties.

[nein09]

@kazarena thank you for your quick reply. I tried Expand() with options.SafeMode = true, as you suggested, and I didn't get any errors. It also gave me all of the properties that are in the original JSON, without dropping any. Here's what I get from that:

[
 {
   "@id": "ad780d9a-e6a7-4a6f-824c-367b40d377b8",
   "@type": [
     "https://schema.org/Dataset"
   ],
   "https://schema.org/citation": [
     {
       "@value": "ACAP Saint John. 2022-04-30. "ACAP Saint John: Temperature Logger Data" (dataset). 2.0.0. DataStream. https://doi.org/10.25976/12cf-1g82."
     }
   ],
   "https://schema.org/creator": [
     {
       "@type": [
         "https://schema.org/Organization"
       ],
       "https://schema.org/name": [
         {
           "@value": "ACAP Saint John"
         }
       ]
     }
   ],
   "https://schema.org/dateModified": [
     {
       "@value": "2022-04-30T15:48:42.840Z"
     }
   ],
   "https://schema.org/datePublished": [
     {
       "@value": "2019-02-04T22:19:58.129Z"
     }
   ],
   "https://schema.org/description": [
     {
       "@value": "ACAP Saint John’s Water Quality Monitoring Programs aim to build upon the community-based water monitoring program established in 1992 in the Greater Saint John Area. The data collected from these various monitoring programs is used to track stressors, determine restoration sites, strengthen/shape policy, and expand community partnerships. Water temperature is logged in various watercourses across the Greater Saint John using temperature logger deployed for months at a time."
     }
   ],
   "https://schema.org/identifier": [
     {
       "@type": [
         "https://schema.org/PropertyValue",
         "datacite:ResourceIdentifier"
       ],
       "datacite:usesIdentifierSchema": [
         {
           "@id": "datacite:doi"
         }
       ],
       "https://schema.org/propertyID": [
         {
           "@value": "DOI"
         }
       ],
       "https://schema.org/url": [
         {
           "@value": "https://doi.org/10.25976/12cf-1g82"
         }
       ],
       "https://schema.org/value": [
         {
           "@value": "10.25976/12cf-1g82"
         }
       ]
     }
   ],
   "https://schema.org/isAccessibleForFree": [
     {
       "@value": true
     }
   ],
   "https://schema.org/keywords": [
     {
       "@value": "Harbour, Saint John, New Brunswick, Water temperature, Urban, Bay of Fundy"
     }
   ],
   "https://schema.org/license": [
     {
       "@value": "https://opendatacommons.org/licenses/by/1-0/"
     }
   ],
   "https://schema.org/measurementTechnique": [
     {
       "@value": "Water temperature is logged using either HOBO or EasyLog temperature loggers attached to cinder blocks and left in the watercourse for various time periods based on the program."
     }
   ],
   "https://schema.org/name": [
     {
       "@value": "ACAP Saint John: Temperature Logger Data"
     }
   ],
   "https://schema.org/publisher": [
     {
       "@type": [
         "https://schema.org/Organization"
       ],
       "https://schema.org/logo": [
         {
           "@value": "https://datastream.org/favicon.svg"
         }
       ],
       "https://schema.org/name": [
         {
           "@value": "DataStream"
         }
       ],
       "https://schema.org/url": [
         {
           "@value": "https://datastream.org"
         }
       ]
     }
   ],
   "https://schema.org/spatialCoverage": [
     {
       "@type": [
         "https://schema.org/Place"
       ],
       "https://schema.org/geo": [
         {
           "@type": [
             "https://schema.org/GeoShape"
           ],
           "https://schema.org/box": [
             {
               "@value": "-66.19002 45.193191 -65.88223 45.40079"
             }
           ]
         }
       ]
     }
   ],
   "https://schema.org/temporalCoverage": [
     {
       "@value": "2016-08-04T13:54:00-03:00/2018-12-06T10:55:50-04:00"
     }
   ],
   "https://schema.org/url": [
     {
       "@value": "https://datastream.org/dataset/ad780d9a-e6a7-4a6f-824c-367b40d377b8"
     }
   ],
   "https://schema.org/variableMeasured": [],
   "https://schema.org/version": [
     {
       "@value": "2.0.0"
     }
   ]
 }
]

Is there a next step that you might suggest? Thanks again.

[kazarena]

@nein09, I think the problem is in the value of the @id property. When I tried using your JSON-LD document on https://json-ld.org/playground/ to generate N-Quads (this is always a good way to understand if there's an issue with the library or the document), I got the same answer as above. But when I tried getting a canonised representation, I got the following error:

jsonld.ValidationError: Safe mode validation error.
{
  "type": [
    "JsonLdEvent"
  ],
  "code": "relative @id reference",
  "level": "warning",
  "message": "Relative @id reference found.",
  "details": {
    "id": "ad780d9a-e6a7-4a6f-824c-367b40d377b8",
    "expandedId": "ad780d9a-e6a7-4a6f-824c-367b40d377b8"
  }
}

The @id property in your document is meant to be an IRI, but it's a string, and there is no @base defined in the context. So it looks like a data modelling problem. For instance, if you add:

"@base": "http://example.com/"

to your context, you will see that RDF gets generated for all the elements.

I hope that helps.

[nein09]

That was exactly it. Thank you so much for your help!