GeoDCAT-AP - How to encode metadata elements on data quality?

Issue: In the INSPIRE and ISO 19115 quality information encoded is as a quality report containing the result of a test (an evaluation) of a given quality measure  according to an evaluation method with a quantitative result (a metric) or a conformance result (pass or fail) as most important outcome. See the website of NOAA for a comprehensive example.

The GeoDCAT-AP draft specification proposes to use the EARL vocabulary to encode data quality. The table below contains an overview of partial syntax bindings.

How would you represent these metadata elements in RDF?

 

PS (Andrea Perego): Mapping tabled revised on 2015-05-18.

In the following table, a ISO data quality report is mapped to class earl:Assertion. The link between the report and the test resource is specified by using earl:subject. The inverse relationship - i.e., the link between the tested resource and the quality report - is specified by using wdrs:describedby.

 

ISO Data Quality Report

Proposed mapping, based on the EARL voc

Measure

Name (O)

earl:assertedBy

earl:Assertor

foaf:name

Identification (O)

dct:identifier

Description (O)

dct:description

Evaluation method (O)

Type (O)

earl:mode

earl:TestMode

earl:TestMode instance name

Description (O)

dct:description

Evaluation procedure (O)

title

earl:test

earl:TestCriterion

dct:title

creation

dct:created

last revision

dct:modified

publication

dct:issued

DateTime (O)

dct:created

Quantitative result (M)

value type (O)

earl:result

earl:TestResult

?

Value unit (M)

?

Statistical error (O)

?

value (M)

earl:outcome

Conformance result (M)

pass (M)

earl:result

earl:TestResult

earl:outcome

explanation (M)

earl:info

specification (M)

title

earl:test earl:TestCriterion

dct:title

creation

dct:created

last revision

dct:modified

publication

dct:issued

Data Quality Scope level (M)    
extent (O)  
level description (C)  

 

Component

Code

Category

feature

Comments

Thu, 28/05/2015 - 07:47

During the 4th WG meeting of 20 May 2015, it was decided by vote that

  • The proposal to provide alignments only for conformance results is accepted.
  • The proposal to use the PROV ontology to represent conformance results is accepted.

There is an explanation of the decision taken in the last version of the GeoDCAT-AP specifications.

 

The GeoDCAT-AP specification only provides a syntax binding for conformity and not for data quality in general. In ISO 19115 conformance and quality information is encoded as a quality report containing the result of a test (an evaluation) of a given quality measure  according to an evaluation method with a quantitative result (a metric) or a conformance result (pass or fail) as most important outcome. For encoding conformance, GeoDCAT-AP proposes to use dct:conformsTo and the W3C Provenance Ontology (PROV-O) as explained in the following paragraphs. For encoding other aspects of data quality, GeoDCAT-AP does not provide a syntax binding as there is a risk that the (future) work of other standards bodies on data quality may make the proposed syntax binding for GeoDCAT-AP outdated. For example, the W3C Data on the Web Best Practices WG is working on a Data Quality standard. To limit the impact, it may be better to only provide a partial mapping for Data Quality / Conformance.

 

DCAT-AP provides a single candidate, dct:conformsTo, which however can be used to map only a conformity of degree ‘conformant’. This is suitable for GeoDCAT-AP Core.

 

Considering how conformity must be expressed in GeoDCAT-AP Extended (see the INSPIRE Metadata Regulation, Part B, §7), possible candidates are the W3C Evaluation and Report Language (EARL) and the W3C Provenance Ontology (PROV-O). The latter candidate was chosen since it would enable wider re-use with respect to the EARL vocabulary, which is more specific, and its use is limited. PROV-O allows encoding conformity as a test activity (prov:Activity) that generated a result encoded with property ‘prov:generated’, corresponding to the degree of conformity, for which the INSPIRE Registry maintains a URI set, see Section 6 of the document. The specification against which the conformance is asserted is encoded via a qualified association (prov:QualifiedAssociation) with a test plan (a prov:Plan) in turn derived from a standard (dct:Standard, also prov:Entity). These associations are made via a chain of properties: ‘prov:qualifiedAssociation’, ‘prov:hadPlan’, and ‘prov:wasDerivedFrom’. 

 

In order to grant interoperability with DCAT-AP, when conformity is of degree “conformant”, the proposal is to use both PROV-O and dct:conformsTo for GeoDCAT-AP Extended.

 

# Resource metadata in GeoDCAT-AP
prov:wasUsedBy [
 a prov:Activity;
# Conformity degree
  prov:generated [
   dct:type <http://inspire.ec.europa.eu/codelist/DegreeOfConformity/conformant> ;
   dct:description “See the referenced specification”@en
  prov:qualifiedAssociation [
   prov:hadPlan [
     a prov:Plan;
     prov:wasDerivedFrom [
# Specification
        a prov:Entity, dct:Standard;
       dct:title "COMMISSION REGULATION (EC) No 976/2009 of 19 October 2009 implementing Directive 2007/2/EC of the European Parliament and of the Council as regards the Network Services"@en
       dct:issued "2009-10-20"^^xsd:date
     ]
   ];
 ];
] .

 

Login or create an account to comment.