Merge pull request #373 from gaiaresources/BDRSPS-1131

BDRSPS-1131 Make datatype attributions named nodes in metadata mapping
gaiaresources · Dec 18, 2024 · 456d013 · 456d013
2 parents dc3ce67 + 8ba2a14
commit 456d013
Show file tree

Hide file tree

Showing 3 changed files with 42 additions and 26 deletions.
diff --git a/abis_mapping/templates/survey_metadata_v3/examples/minimal.ttl b/abis_mapping/templates/survey_metadata_v3/examples/minimal.ttl
@@ -52,21 +52,11 @@
 
 <https://linked.data.gov.au/dataset/bdr/datatypes/surveyID/CSIRO> a rdfs:Datatype ;
     skos:prefLabel "surveyID source" ;
-    prov:qualifiedAttribution [ a prov:Attribution ;
-            prov:agent <https://linked.data.gov.au/dataset/bdr/orgs/CSIRO> ;
-            prov:hadRole <https://linked.data.gov.au/def/data-roles/principalInvestigator> ],
-        [ a prov:Attribution ;
-            prov:agent <https://linked.data.gov.au/dataset/bdr/orgs/CSIRO> ;
-            prov:hadRole <https://linked.data.gov.au/def/data-roles/principalInvestigator> ] .
+    prov:qualifiedAttribution <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/attribution/CSIRO/principalInvestigator> .
 
 <https://linked.data.gov.au/dataset/bdr/datatypes/surveyID/NSW-Department-of-Planning-Industry-and-Environment> a rdfs:Datatype ;
     skos:prefLabel "surveyID source" ;
-    prov:qualifiedAttribution [ a prov:Attribution ;
-            prov:agent <https://linked.data.gov.au/dataset/bdr/orgs/NSW-Department-of-Planning-Industry-and-Environment> ;
-            prov:hadRole <https://linked.data.gov.au/def/data-roles/principalInvestigator> ],
-        [ a prov:Attribution ;
-            prov:agent <https://linked.data.gov.au/dataset/bdr/orgs/NSW-Department-of-Planning-Industry-and-Environment> ;
-            prov:hadRole <https://linked.data.gov.au/def/data-roles/principalInvestigator> ] .
+    prov:qualifiedAttribution <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/attribution/NSW-Department-of-Planning-Industry-and-Environment/principalInvestigator> .
 
 <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/attribute/surveyType/Wet-pitfall-trapping> a tern:Attribute ;
     schema:isPartOf <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000> ;
@@ -92,6 +82,14 @@
     tern:hasSimpleValue "Insecta" ;
     tern:hasValue <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/value/targetTaxonomicScope/Insecta> .
 
+<https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/attribution/CSIRO/principalInvestigator> a prov:Attribution ;
+    prov:agent <https://linked.data.gov.au/dataset/bdr/orgs/CSIRO> ;
+    prov:hadRole <https://linked.data.gov.au/def/data-roles/principalInvestigator> .
+
+<https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/attribution/NSW-Department-of-Planning-Industry-and-Environment/principalInvestigator> a prov:Attribution ;
+    prov:agent <https://linked.data.gov.au/dataset/bdr/orgs/NSW-Department-of-Planning-Industry-and-Environment> ;
+    prov:hadRole <https://linked.data.gov.au/def/data-roles/principalInvestigator> .
+
 <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/bdr-cv/attribute/targetTaxonomicScope/Coleoptera> a skos:Concept ;
     skos:broader <https://linked.data.gov.au/def/nrm/7ea12fed-6b87-4c20-9ab4-600b32ce15ec> ;
     skos:definition "A type of targetTaxonomicScope" ;
@@ -126,6 +124,12 @@
     rdfs:label "Insecta" ;
     rdf:value <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/bdr-cv/attribute/targetTaxonomicScope/Insecta> .
 
+<https://linked.data.gov.au/dataset/bdr/orgs/CSIRO> a prov:Agent ;
+    schema:name "CSIRO" .
+
+<https://linked.data.gov.au/dataset/bdr/orgs/NSW-Department-of-Planning-Industry-and-Environment> a prov:Agent ;
+    schema:name "NSW Department of Planning, Industry and Environment" .
+
 <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/Survey/COL1> a tern:Survey ;
     bdr:purpose "Summer sampling for peak insect diversity." ;
     bdr:target "Coleoptera",
@@ -172,12 +176,6 @@
         "woodland" ;
     schema:name "Disentangling the effects of farmland use, habitat edges, and vegetation structure on ground beetle morphological traits - Winter" .
 
-<https://linked.data.gov.au/dataset/bdr/orgs/CSIRO> a prov:Agent ;
-    schema:name "CSIRO" .
-
-<https://linked.data.gov.au/dataset/bdr/orgs/NSW-Department-of-Planning-Industry-and-Environment> a prov:Agent ;
-    schema:name "NSW Department of Planning, Industry and Environment" .
-
 <https://linked.data.gov.au/dataset/bdr/00000000-0000-0000-0000-000000000000/survey/plan/COL1> a prov:Plan ;
     schema:citation "Ng, K., Barton, P.S., Blanchard, W. et al. Disentangling the effects of farmland use, habitat edges, and vegetation structure on ground beetle morphological traits. Oecologia 188, 645–657 (2018). https://doi.org/10.1007/s00442-018-4180-9\"" ;
     schema:description "Our experimental design consisted of four 400 m transects running from inside each woodland patch out into four adjoining farmland uses (crop, rested, woody debris application, revegetation plantings). To quantify potential edge efects on beetle species traits, we sampled beetles at five locations along each transect: 200 and 20 m inside woodlands, 200 and 20 m inside farmlands, and at the woodland–farmland edge (0 m). Each sampling location comprised a pair of wet invertebrate pitfall traps.  separated by a drift fence (60 cm long x 10 cm high) to help direct arthropods into traps. We opened a total of 220 pairs of traps for 14 days during spring (Oct–Nov 2014), and repeated sampling during summer (January–February 2015). Beetle samples from each pitfall trap pair, and across the two time periods, were pooled to provide one sample per sampling location." ;

diff --git a/abis_mapping/templates/survey_metadata_v3/mapping.py b/abis_mapping/templates/survey_metadata_v3/mapping.py
@@ -35,6 +35,7 @@ class SurveyIDDatatype:
 
     name: str
     datatype: rdflib.URIRef
+    attribution: rdflib.URIRef
     agent: rdflib.URIRef
 
 
@@ -222,6 +223,7 @@ def apply_mapping_row(
                     SurveyIDDatatype(
                         name=raw_org,
                         datatype=utils.iri_patterns.datatype_iri("surveyID", raw_org),
+                        attribution=utils.iri_patterns.attribution_iri(base_iri, "principalInvestigator", raw_org),
                         agent=utils.iri_patterns.agent_iri(raw_org),
                     )
                 )
@@ -262,7 +264,15 @@ def apply_mapping_row(
             # Add survey ID source datatype nodes
             self.add_survey_id_source_datatypes(
                 uri=so_obj.datatype,
+                attribution=so_obj.attribution,
+                graph=graph,
+            )
+
+            # Add attribution
+            self.add_attribution(
+                uri=so_obj.attribution,
                 agent=so_obj.agent,
+                role=PRINCIPAL_INVESTIGATOR,
                 graph=graph,
             )
 
@@ -541,29 +551,37 @@ def add_temporal_coverage(
     def add_survey_id_source_datatypes(
         self,
         uri: rdflib.URIRef,
-        agent: rdflib.URIRef,
+        attribution: rdflib.URIRef,
         graph: rdflib.Graph,
     ) -> None:
         """Adds the source datatype nodes to graph.
 
         Args:
             uri (rdflib.URIRef): The reference uri.
-            agent (rdflib.URIRef): Agent uri.
+            attribution (rdflib.URIRef): Attribution uri.
             graph (rdflib.Graph): Graph to be modified.
         """
         # Add type
         graph.add((uri, a, rdflib.RDFS.Datatype))
 
         # Add label
         graph.add((uri, rdflib.SKOS.prefLabel, rdflib.Literal("surveyID source")))
-
         # Add attribution
-        attribution = rdflib.BNode()
-        graph.add((attribution, a, rdflib.PROV.Attribution))
-        graph.add((attribution, rdflib.PROV.agent, agent))
-        graph.add((attribution, rdflib.PROV.hadRole, PRINCIPAL_INVESTIGATOR))
         graph.add((uri, rdflib.PROV.qualifiedAttribution, attribution))
 
+    def add_attribution(
+        self,
+        uri: rdflib.URIRef,
+        agent: rdflib.URIRef,
+        role: rdflib.URIRef,
+        graph: rdflib.Graph,
+    ) -> None:
+        """Add the prov:Attribution nodes to the graph."""
+        # Add attribution
+        graph.add((uri, a, rdflib.PROV.Attribution))
+        graph.add((uri, rdflib.PROV.agent, agent))
+        graph.add((uri, rdflib.PROV.hadRole, role))
+
     def add_agent(
         self,
         uri: rdflib.URIRef,

diff --git a/abis_mapping/utils/iri_patterns.py b/abis_mapping/utils/iri_patterns.py
@@ -322,7 +322,7 @@ def plan_iri(
 
 def attribution_iri(
     base_iri: rdflib.Namespace,
-    role: Literal["resourceProvider", "owner", "rightsHolder", "creator"],
+    role: Literal["resourceProvider", "owner", "rightsHolder", "creator", "principalInvestigator"],
     source: str,
 ) -> rdflib.URIRef:
     """Get the IRI to use for a prov:Attribution node.