Feature/sparql qc queries

.github/deployment/sparql/CCO_classes_have_BFO_superclass.sparql

@@ -13,7 +13,7 @@ PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
 SELECT DISTINCT ?class ?label
 WHERE {
 ?class a owl:Class .
-FILTER (regex(str(?class),"http://www.ontologyrepository.com/CommonCoreOntologies/"))
+FILTER (regex(str(?class),"https://www.commoncoreontologies.org/"))
 FILTER NOT EXISTS {?class rdfs:subClassOf+ <http://purl.obolibrary.org/obo/BFO_0000001> }
 OPTIONAL {?class rdfs:label ?label}
 }

.github/deployment/sparql/annotation_language_tag.sparql

@@ -0,0 +1,60 @@
+# Title:
+#    Annotation Values Must Have English Language Tag
+# Constraint Description:
+#    All natural-language annotation values on CCO ontology elements must carry
+#    an English (@en) language tag. This applies to: rdfs:label, skos:definition,
+#    rdfs:comment, skos:altLabel, skos:prefLabel, skos:editorNote,
+#    cco:ont00001737 [doctrinal definition], and cco:ont00001748 [doctrinal label].
+#    Values typed as xsd:anyURI are exempt (they are IRI references, not NL text).
+#    English dialect tags such as @en-US and @en-GB are accepted.
+# Severity:
+#    Error
+# Reference:
+#   # Author:
+#    github.com/shanmukhkalasamudram
+
+PREFIX owl:  <http://www.w3.org/2002/07/owl#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
+PREFIX xsd:  <http://www.w3.org/2001/XMLSchema#>
+PREFIX cco:  <https://www.commoncoreontologies.org/>
+
+SELECT DISTINCT ?resource ?prop ?val ?error
+WHERE {
+  VALUES ?type {
+    owl:Class
+    owl:ObjectProperty
+    owl:AnnotationProperty
+    owl:DatatypeProperty
+    owl:NamedIndividual
+  }
+  # Scope to natural-language annotation properties only.
+  # xsd:anyURI-typed values (IRI references) are handled by the datatype filter below.
+  VALUES ?prop {
+    rdfs:label
+    skos:definition
+    rdfs:comment
+    skos:altLabel
+    skos:prefLabel
+    skos:editorNote
+    cco:ont00001737
+    cco:ont00001748
+  }
+  ?resource a ?type .
+  FILTER (STRSTARTS(STR(?resource), "https://www.commoncoreontologies.org/"))
+  FILTER (!isBlank(?resource))
+  ?resource ?prop ?val .
+  FILTER (isLiteral(?val))
+  # langMatches handles dialect tags: @en-US, @en-GB all pass.
+  # lang("") of a plain literal returns "" — langMatches("", "en") = FALSE — flagged.
+  FILTER (!langMatches(lang(?val), "en"))
+  # Allow xsd:anyURI values (IRI references do not need a language tag).
+  FILTER (datatype(?val) != xsd:anyURI)
+  BIND (CONCAT(
+    "ERROR: Term ", STR(?resource),
+    " has annotation <", STR(?prop),
+    "> with value \"", STR(?val),
+    "\" that is missing an @en language tag (lang='", lang(?val), "')."
+  ) AS ?error)
+}
+ORDER BY ?resource ?prop

.github/deployment/sparql/curated_in_matches_ontology.sparql

@@ -0,0 +1,58 @@
+# Title:
+#    cco:ont00001760 Value Must Point to a Declared CCO Ontology
+# Constraint Description:
+#    The value of cco:ont00001760 [is curated in ontology] on every CCO ontology
+#    element must be the IRI of a CCO ontology that is actually declared as
+#    owl:Ontology within the CCO namespace
+#    (https://www.commoncoreontologies.org/).
+#    This catches values that:
+#      - Point to a non-existent or deleted ontology module.
+#      - Use the old v1 namespace (http://www.ontologyrepository.com/...).
+#      - Contain a typo in the ontology name.
+# Severity:
+#    Error
+# Reference:
+#    CCO Release Process v2, Step 6c / Step 6f
+# Author:
+#    github.com/shanmukhkalasamudram
+
+PREFIX owl:  <http://www.w3.org/2002/07/owl#>
+PREFIX cco:  <https://www.commoncoreontologies.org/>
+
+SELECT DISTINCT ?resource ?curatedIn ?error
+WHERE {
+  VALUES ?type {
+    owl:Class
+    owl:ObjectProperty
+    owl:AnnotationProperty
+    owl:DatatypeProperty
+    owl:NamedIndividual
+  }
+  ?resource a ?type .
+  FILTER (STRSTARTS(STR(?resource), "https://www.commoncoreontologies.org/"))
+  FILTER (!isBlank(?resource))
+  ?resource cco:ont00001760 ?curatedIn .
+  # The merged file only carries one owl:Ontology IRI (CommonCoreOntologiesMerged),
+  # so dynamic lookup against owl:Ontology triples is not feasible for merged runs.
+  # Instead validate against the fixed allowlist of the 11 CCO module ontology IRIs.
+  # Update this list whenever a module is added or removed.
+  FILTER (STR(?curatedIn) NOT IN (
+    "https://www.commoncoreontologies.org/AgentOntology",
+    "https://www.commoncoreontologies.org/ArtifactOntology",
+    "https://www.commoncoreontologies.org/CurrencyUnitOntology",
+    "https://www.commoncoreontologies.org/EventOntology",
+    "https://www.commoncoreontologies.org/ExtendedRelationOntology",
+    "https://www.commoncoreontologies.org/FacilityOntology",
+    "https://www.commoncoreontologies.org/GeospatialOntology",
+    "https://www.commoncoreontologies.org/InformationEntityOntology",
+    "https://www.commoncoreontologies.org/QualityOntology",
+    "https://www.commoncoreontologies.org/TimeOntology",
+    "https://www.commoncoreontologies.org/UnitsOfMeasureOntology"
+  ))
+  BIND (CONCAT(
+    "ERROR: Term <", STR(?resource),
+    "> has cco:ont00001760 value <", STR(?curatedIn),
+    "> which does not match any known CCO module ontology IRI."
+  ) AS ?error)
+}
+ORDER BY ?resource

.github/deployment/sparql/duplicate_iri_number.sparql

@@ -0,0 +1,42 @@
+# duplicate_iri_number.sparql
+# Step 6 QC — Flags any two OWL entities whose IRI ends in the same
+# "ont" + 8-digit numeric fragment, regardless of namespace.
+#
+# Why this matters:
+#   The canonical IRI pattern is https://www.commoncoreontologies.org/ont<8digits>
+#   If the same 8-digit number appears under a different base (e.g., old v1
+#   namespace http://www.ontologyrepository.com/CommonCoreOntologies/ or a
+#   typo namespace), two distinct IRIs share the same logical ID — creating
+#   ambiguity, potential reasoner issues, and broken cross-references.
+#
+# Reports: pairs of IRIs sharing the same numeric fragment, plus that fragment.
+# Expected result on a clean release: 0 violations.
+
+PREFIX owl: <http://www.w3.org/2002/07/owl#>
+PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
+
+SELECT DISTINCT ?iri1 ?iri2 ?sharedID WHERE {
+    VALUES ?owlType {
+        owl:Class
+        owl:ObjectProperty
+        owl:AnnotationProperty
+        owl:DatatypeProperty
+        owl:NamedIndividual
+    }
+
+    # Both IRIs must be declared OWL entities with the ont+8digit pattern
+    ?iri1 rdf:type ?owlType .
+    ?iri2 rdf:type ?owlType .
+
+    FILTER(?iri1 != ?iri2)
+    FILTER(REGEX(STR(?iri1), "ont[0-9]{8}$"))
+    FILTER(REGEX(STR(?iri2), "ont[0-9]{8}$"))
+
+    # Extract the shared numeric fragment (e.g. "ont00001234")
+    BIND(REPLACE(STR(?iri1), "^.*(ont[0-9]{8})$", "$1") AS ?sharedID)
+    FILTER(STRENDS(STR(?iri2), ?sharedID))
+
+    # Report each pair once only
+    FILTER(STR(?iri1) < STR(?iri2))
+}
+ORDER BY ?sharedID

.github/deployment/sparql/exactly_1_curated_in.sparql

@@ -0,0 +1,56 @@
+# Title:
+#    Exactly One cco:ont00001760 [is curated in ontology] Per Ontology Element
+# Constraint Description:
+#    All CCO ontology elements (classes, properties, individuals) must have
+#    exactly one cco:ont00001760 annotation. This query flags:
+#      (a) Elements with NO cco:ont00001760 annotation at all.
+#      (b) Elements with MORE THAN ONE cco:ont00001760 annotation.
+# Severity:
+#    Error
+# Reference:
+#    CCO Release Process v2, Step 6c
+# Author:
+#    github.com/shanmukhkalasamudram
+
+PREFIX owl:  <http://www.w3.org/2002/07/owl#>
+PREFIX cco:  <https://www.commoncoreontologies.org/>
+
+SELECT DISTINCT ?resource ?error
+WHERE {
+  {
+    # Branch (a): missing cco:ont00001760 entirely
+    # All patterns are self-contained inside each UNION branch so that
+    # ?resource is properly bound before FILTER NOT EXISTS is evaluated.
+    VALUES ?type {
+      owl:Class
+      owl:ObjectProperty
+      owl:AnnotationProperty
+      owl:DatatypeProperty
+      owl:NamedIndividual
+    }
+    ?resource a ?type .
+    FILTER (regex(str(?resource), "https://www.commoncoreontologies.org/"))
+    FILTER (!isBlank(?resource))
+    FILTER NOT EXISTS { ?resource cco:ont00001760 ?any }
+    BIND (CONCAT("ERROR: Term ", str(?resource), " is missing cco:ont00001760 [is curated in ontology].") AS ?error)
+  }
+  UNION
+  {
+    # Branch (b): more than one cco:ont00001760 annotation
+    VALUES ?type {
+      owl:Class
+      owl:ObjectProperty
+      owl:AnnotationProperty
+      owl:DatatypeProperty
+      owl:NamedIndividual
+    }
+    ?resource a ?type .
+    FILTER (regex(str(?resource), "https://www.commoncoreontologies.org/"))
+    FILTER (!isBlank(?resource))
+    ?resource cco:ont00001760 ?val1 .
+    ?resource cco:ont00001760 ?val2 .
+    FILTER (STR(?val1) != STR(?val2))
+    BIND (CONCAT("ERROR: Term ", str(?resource), " has duplicate cco:ont00001760 values: <", str(?val1), "> and <", str(?val2), ">.") AS ?error)
+  }
+}
+ORDER BY ?resource

.github/deployment/sparql/exactly_1_label.sparql

@@ -0,0 +1,59 @@
+# Title:
+#    Exactly One rdfs:label Per Language Per Ontology Element
+# Constraint Description:
+#    All CCO ontology elements (classes, properties, individuals) must have
+#    exactly one rdfs:label. This query flags:
+#      (a) Elements with NO rdfs:label at all.
+#      (b) Elements with MORE THAN ONE rdfs:label in the same language.
+# Severity:
+#    Error
+# Reference:
+#    CCO Release Process v2, Step 6c
+#    Pattern based on: exactly_1_prefLabel_per_lang.sparql
+# Author:
+#    github.com/shanmukhkalasamudram
+
+PREFIX owl:  <http://www.w3.org/2002/07/owl#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+SELECT DISTINCT ?resource ?error
+WHERE {
+  {
+    # Branch (a): missing rdfs:label entirely
+    # All patterns must be self-contained inside each UNION branch so that
+    # ?resource is properly bound before FILTER NOT EXISTS is evaluated.
+    VALUES ?type {
+      owl:Class
+      owl:ObjectProperty
+      owl:AnnotationProperty
+      owl:DatatypeProperty
+      owl:NamedIndividual
+    }
+    ?resource a ?type .
+    FILTER (regex(str(?resource), "https://www.commoncoreontologies.org/"))
+    FILTER (!isBlank(?resource))
+    FILTER NOT EXISTS { ?resource rdfs:label ?anyLbl }
+    BIND (CONCAT("ERROR: Term ", str(?resource), " has no rdfs:label.") AS ?error)
+  }
+  UNION
+  {
+    # Branch (b): more than one rdfs:label in the same language
+    VALUES ?type {
+      owl:Class
+      owl:ObjectProperty
+      owl:AnnotationProperty
+      owl:DatatypeProperty
+      owl:NamedIndividual
+    }
+    ?resource a ?type .
+    FILTER (regex(str(?resource), "https://www.commoncoreontologies.org/"))
+    FILTER (!isBlank(?resource))
+    ?resource rdfs:label ?lbl1 .
+    ?resource rdfs:label ?lbl2 .
+    FILTER (?lbl1 != ?lbl2)
+    FILTER (lang(?lbl1) = lang(?lbl2))
+    FILTER (lang(?lbl1) != "")
+    BIND (CONCAT("ERROR: Term ", str(?resource), " has duplicate rdfs:label values in language '", lang(?lbl1), "': \"", str(?lbl1), "\" and \"", str(?lbl2), "\".") AS ?error)
+  }
+}
+ORDER BY ?resource

.github/deployment/sparql/iri_format_check.sparql

@@ -0,0 +1,38 @@
+# Title:
+#    CCO Term IRI Format Check
+# Constraint Description:
+#    All CCO ontology elements (classes, properties, individuals) must have an IRI
+#    matching the canonical CCO format:
+#      https://www.commoncoreontologies.org/ont<8 digits>
+#    Example: https://www.commoncoreontologies.org/ont00001760
+#    Ontology-level IRIs (owl:Ontology) are not term IRIs and are intentionally
+#    excluded from this check.
+# Severity:
+#    Error
+# Reference:
+#    CCO Release Process v2, Step 6e
+# Author:
+#    github.com/shanmukhkalasamudram
+
+PREFIX owl:  <http://www.w3.org/2002/07/owl#>
+
+SELECT DISTINCT ?resource ?error
+WHERE {
+  VALUES ?type {
+    owl:Class
+    owl:ObjectProperty
+    owl:AnnotationProperty
+    owl:DatatypeProperty
+    owl:NamedIndividual
+  }
+  ?resource a ?type .
+  FILTER (STRSTARTS(STR(?resource), "https://www.commoncoreontologies.org/"))
+  FILTER (!isBlank(?resource))
+  # Flag any term IRI that does NOT match ont + exactly 8 digits.
+  FILTER (!REGEX(STR(?resource), "^https://www\\.commoncoreontologies\\.org/ont[0-9]{8}$"))
+  BIND (CONCAT(
+    "ERROR: Term IRI <", STR(?resource),
+    "> does not match the required CCO format https://www.commoncoreontologies.org/ont<8 digits>."
+  ) AS ?error)
+}
+ORDER BY ?resource

.github/deployment/sparql/missing_definition_source.sparql

@@ -0,0 +1,41 @@
+# Title:
+#    Definition Source Required
+# Constraint Description:
+#    All CCO classes and object properties that have a skos:definition should also
+#    have a cco:ont00001754 [definition source] annotation. Terms missing a
+#    definition source are flagged as warnings. Some terms may legitimately lack a
+#    source (e.g., primitive relations); those should be reviewed case by case.
+# Severity:
+#    Warning
+# Reference:
+#    CCO Release Process v2, Step 6h
+#    Pattern based on: min_1_eng_def.sparql
+# Author:
+#    github.com/shanmukhkalasamudram
+
+PREFIX owl:  <http://www.w3.org/2002/07/owl#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
+PREFIX cco:  <https://www.commoncoreontologies.org/>
+
+SELECT DISTINCT ?resource ?label ?error
+WHERE {
+  VALUES ?type { owl:Class owl:ObjectProperty }
+  ?resource a ?type .
+
+  # Must have a definition (already caught by min_1_eng_def if missing)
+  ?resource skos:definition ?definition .
+
+  # Scope to current CCO terms only
+  FILTER (regex(str(?resource), "https://www.commoncoreontologies.org/"))
+  FILTER (!isBlank(?resource))
+
+  # Flag those missing a definition source
+  OPTIONAL { ?resource cco:ont00001754 ?source }
+  FILTER (!bound(?source))
+
+  OPTIONAL { ?resource rdfs:label ?label }
+
+  BIND (CONCAT("WARNING: Term ", str(?resource), " has a definition but is missing cco:ont00001754 [definition source].") AS ?error)
+}
+ORDER BY ?resource

.github/deployment/sparql/no_cco_elucidation.sparql

@@ -0,0 +1,28 @@
+# Title:
+#    No Use of Deprecated cco:elucidation Property
+# Constraint Description:
+#    The annotation property cco:elucidation (from the old CCO v1 namespace) is
+#    deprecated. It was replaced by skos:editorNote in CCO v2. Any CCO term that
+#    still uses cco:elucidation as a predicate must be updated.
+# Severity:
+#    Error
+# Author:
+#	 github.com/shanmukhkalasamudram
+#    Inspired by: Giacomo Decolle
+
+PREFIX owl:  <http://www.w3.org/2002/07/owl#>
+PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
+
+SELECT DISTINCT ?subject ?label ?error
+WHERE {
+  ?subject <http://www.ontologyrepository.com/CommonCoreOntologies/elucidation> ?value .
+
+  # Scope to current CCO terms only
+  FILTER (regex(str(?subject), "https://www.commoncoreontologies.org/"))
+  FILTER (!isBlank(?subject))
+
+  OPTIONAL { ?subject rdfs:label ?label }
+
+  BIND (CONCAT("ERROR: Term ", str(?subject), " uses deprecated cco:elucidation. Replace with skos:editorNote.") AS ?error)
+}
+ORDER BY ?subject