### Base SHACL ### ### Fundamental constraint shapes for the Martialis database ### ### Frederico Munoz , Martialis Project ### ### Licence: CC0-1.0 (public domain dedication) ### Attribution requested but not required: cite "Martialis — https://martial.is" ### NB: "every Assertion must have at least one source" (and similar) ### is enforced in martialis-quality.shacl.ttl as a quality warning, ### not a structural constraint. Structural SHACL covers referential ### integrity and semantic correctness; data completeness requirements ### belong in the quality tier. The one exception is NegativeAssertion: ### an explicit denial without sources is worse than no data, so its ### evidence requirement is structural. ### ### Shapes target the concrete classes stamped by the conversion ### pipeline (bin/martialis rdf), so validation runs without any ### RDFS/OWL inference and without loading the ontology graph. ### ### Enum membership: concept-scheme checks (added in the SKOS vocabulary ### migration, ontology 2.0.0). Before this migration, enum-valued ### properties were plain xsd:string literals, and JSON Schema alone ### enforced their legal values - a second sh:in copy in SHACL would ### have had no additional referential or type-integrity value, so the ### policy was deliberately not to duplicate it (see this file's git ### history for that original reasoning). ### Since the migration, most of those same properties hold concept ### IRIs from ontology/vocabularies/*.ttl instead, and checking that an ### IRI resolves to a skos:Concept in the RIGHT scheme is exactly the ### kind of referential integrity JSON Schema structurally cannot ### express (it only ever sees the enum string, never the graph). Where ### that applies, the property shape below carries: ### sh:class skos:Concept ; ### sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; ### A handful of properties were deliberately EXCLUDED from the IRI ### migration under the SKOS plan's governing rule ("concept IRIs for ### vocabularies Martialis defines, literals for tokens defined by an ### external standard") - sourceType/sourceLocatorLabel/contributorRole ### (CSL), peerReview (DRIVER/OpenAIRE), and confidenceLevel (an ordinal ### integer, not a vocabulary). Those keep their pre-migration ### unenforced-enum or sh:datatype xsd:string shapes; the comment on ### each says so explicitly rather than leaving it to be inferred. @prefix sh: . @prefix rdf: . @prefix rdfs: . @prefix xsd: . @prefix martialis: . @prefix shapes: . @prefix schema: . @prefix prov: . @prefix org: . @prefix owl: . @prefix dcterms: . @prefix vann: . @prefix skos: . # Shapes graph self-description ################################################################# # Declared for documentation generators (pyLODE valpub) and for # self-description; pyshacl does not follow owl:imports, and # bin/martialis loads this file directly, so neither the ontology # import nor these annotation triples affect validation. a owl:Ontology , sh:ShapesGraph ; dcterms:title "Martialis Structural Constraints"@en ; dcterms:description "Fundamental constraint shapes for the Martialis database: referential integrity and semantic correctness. This is the blocking validation tier; data-completeness requirements belong to the quality tier (martialis-quality.shacl.ttl). Shapes target the concrete classes stamped by the conversion pipeline (bin/martialis rdf), so validation runs without RDFS/OWL inference and without loading the ontology graph."@en ; dcterms:license ; dcterms:created "2026-02-28"^^xsd:date ; dcterms:modified "2026-09-07"^^xsd:date ; owl:versionInfo "1.2.0" ; owl:imports ; vann:preferredNamespacePrefix "shapes" ; vann:preferredNamespaceUri ; rdfs:seeAlso . # Person ################################################################# shapes:PersonShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Person"@en ; rdfs:comment "Core Person integrity: required name, well-formed birth/death life events, and referential integrity of every assertion link."@en ; sh:targetClass martialis:Person ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:message "A Person must have at least one name." ] ; sh:property [ sh:severity sh:Violation ; sh:path schema:alternateName ; sh:datatype rdf:langString ; sh:message "Alternate names must be language-tagged string literals." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasTeacherAssertion ; sh:class martialis:TeacherAssertion ; sh:message "teachers must be TeacherAssertion nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasNegativeAssertion ; sh:class martialis:NegativeAssertion ; sh:message "negated teacher entries must be NegativeAssertion nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasContextualTrainingAssertion ; sh:class martialis:ContextualTrainingAssertion ; sh:message "teacher_id: null (contextual training) entries must be ContextualTrainingAssertion nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasRankAssertion ; sh:class martialis:RankAssertion ; sh:message "rank must be expressed via RankAssertion." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasHonorAssertion ; sh:class martialis:HonorAssertion ; sh:message "honors must be expressed via HonorAssertion." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasDocumentTransmissionAssertion ; sh:class martialis:DocumentTransmissionAssertion ; sh:message "learned_from entries must be DocumentTransmissionAssertion nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasIdentityAssertion ; sh:class martialis:IdentityAssertion ; sh:message "possibly_same_as entries must be IdentityAssertion nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:taughtAt ; sh:class martialis:Place ; sh:message "taughtAt (materialized from teacher interactions) references a non-existent place." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:trainedAt ; sh:class martialis:Place ; sh:message "trainedAt (materialized from teacher/contextual training interactions) references a non-existent place." ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "person birth after death" ; sh:message "Person's earliest possible birth year is after their latest possible death year." ; sh:select """ PREFIX martialis: PREFIX xsd: SELECT $this ?birthEarliest ?deathLatest WHERE { $this martialis:birthEvent/martialis:earliestYear ?birthEarliest . $this martialis:deathEvent/martialis:latestYear ?deathLatest . FILTER(xsd:integer(STR(?birthEarliest)) > xsd:integer(STR(?deathLatest))) } """ ] . # TeacherAssertion ################################################################# shapes:TeacherAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "TeacherAssertion"@en ; rdfs:comment "Every teacher claim needs a subject, an object teacher, and at least one interaction."@en ; sh:targetClass martialis:TeacherAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "TeacherAssertion subject must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "TeacherAssertion object must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:assertedStyle ; sh:class martialis:Style ; sh:message "Teacher assertion references non-existent style." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasInteraction ; sh:minCount 1 ; sh:class martialis:TeacherInteraction ; sh:message "Teacher interactions must be TeacherInteraction nodes; every TeacherAssertion needs at least one." ] ; # role is optional on a TeacherAssertion (JSON Schema governs the # YAML enum), but when present its exported concept IRI must be in # the teacher-role scheme - the mirror of ParticipationShape's # event-role check on the other side of the martialis:role split. sh:property [ sh:severity sh:Violation ; sh:path martialis:role ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "teachers[].role must be a teacher-role concept (student, uchi-deshi, seminar, private, examination, coach)." ] . # NegativeAssertion ################################################################# shapes:NegativeAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "NegativeAssertion"@en ; rdfs:comment "Denials are structurally required to carry evidence: an explicit denial without sources is worse than no data."@en ; sh:targetClass martialis:NegativeAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "NegativeAssertion subject must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "NegativeAssertion object must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:assertedStyle ; sh:class martialis:Style ; sh:message "Negative assertion references non-existent style." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:message "A NegativeAssertion (explicit denial) must cite at least one source." ] . # ContextualTrainingAssertion ################################################################# shapes:ContextualTrainingAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "ContextualTrainingAssertion"@en ; rdfs:comment "Training with no identified teacher: a subject, a style, and at least one interaction, each anchored to a Place or Event. No object - that is what distinguishes it from a TeacherAssertion, and object is structurally forbidden here so the class cannot drift into an unsourced lineage claim."@en ; sh:targetClass martialis:ContextualTrainingAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "ContextualTrainingAssertion subject must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:maxCount 0 ; sh:message "ContextualTrainingAssertion must not have an object - the teacher is unidentified by construction. A claim with a known teacher belongs in teachers[] with a real teacher_id instead." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:assertedStyle ; sh:minCount 1 ; sh:maxCount 1 ; sh:class martialis:Style ; sh:message "ContextualTrainingAssertion requires exactly one style." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasInteraction ; sh:minCount 1 ; sh:class martialis:TeacherInteraction ; sh:message "Contextual training interactions must be TeacherInteraction nodes; every ContextualTrainingAssertion needs at least one." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:role ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "contextual_training[].role must be a teacher-role concept (student, uchi-deshi, seminar, private, examination, coach)." ] ; # The anchor rule: JSON Schema enforces it on authoring (place_id or # event_ids required per interaction when teacher_id is null), but it # is re-checked here as the structural guarantee for any RDF that # reaches this class by another route (hand-built graphs, a future # second producer) - the same belt-and-braces stance the project # takes for "sources required" elsewhere. sh:sparql [ a sh:SPARQLConstraint ; sh:name "contextual training interaction unanchored" ; sh:message "A ContextualTrainingAssertion interaction must reference a Place (martialis:interactionPlace) or an Event (martialis:atEvent) - a training claim with no identified teacher and no anchor asserts nothing checkable." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:hasInteraction ?i . FILTER NOT EXISTS { ?i martialis:interactionPlace ?p } FILTER NOT EXISTS { ?i martialis:atEvent ?e } } """ ] . # TeacherInteraction ################################################################# shapes:TeacherInteractionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "TeacherInteraction"@en ; rdfs:comment "Interaction records must reference a Place and a well-formed period."@en ; sh:targetClass martialis:TeacherInteraction ; sh:property [ sh:severity sh:Violation ; sh:path martialis:interactionPlace ; sh:maxCount 1 ; sh:class martialis:Place ; sh:message "Interaction place must reference a Place." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:interactionPeriod ; sh:maxCount 1 ; sh:class martialis:TimeInterval ; sh:message "Interaction period must be a TimeInterval." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:atEvent ; sh:class martialis:Event ; sh:message "Interaction event_ids reference non-existent event." ] . # RankAssertion ################################################################# shapes:RankAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "RankAssertion"@en ; rdfs:comment "Rank claims: subject, style object, and rankSystem required; grantedBy (a Person) and assertedOrg (an Organization) are distinct granting routes. rankOrdinal and rankGrade are never both set (nonsensical - a grade can't be both numbered and enumerated) - see the sh:not below. Neither set is legitimate: the direction (rankType) may be documented without the source giving a specific degree (e.g. a source that only says 'black belt' without a dan number)."@en ; sh:targetClass martialis:RankAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "RankAssertion subject must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:minCount 1 ; sh:class martialis:Style ; sh:message "RankAssertion object must be a Style." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:grantedBy ; sh:class martialis:Person ; sh:message "Rank granted_by references non-existent person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:assertedOrg ; sh:class martialis:Organization ; sh:message "Rank assertion references non-existent organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:rankSystem ; sh:minCount 1 ; sh:maxCount 1 ; sh:class martialis:RankingSystem ; sh:message "rank_system_id must reference an existing RankingSystem entity." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:rankType ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "rank_type must be a rank-type concept (dan or kyu)." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:rankOrdinal ; sh:maxCount 1 ; sh:datatype xsd:integer ; sh:message "rank_ordinal must be an integer." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:rankGrade ; sh:maxCount 1 ; sh:class skos:Concept ; # Checks membership in SOME enumerated rank-grade scheme, not # that the grade belongs to the SPECIFIC RankingSystem this # assertion's rankSystem names - that cross-reference (join # through the referenced entity's rdfs:seeAlso to find its # declared grade_scheme) is not expressible as a simple sh:in/ # sh:hasValue and would need sh:sparql; left as a documented gap # rather than solved here - see person.schema.json's Rank.rank_grade # comment for the same caveat on the JSON Schema side. Extend the # sh:or list below when a new enumerated scheme is added. sh:node [ sh:or ( [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ) ] ; sh:message "rank_grade must be a concept in a known enumerated rank-grade scheme." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:atEvent ; sh:class martialis:Event ; sh:message "Rank event_ids reference non-existent event." ] . # rankOrdinal and rankGrade are NOT mutually exclusive: BJJ_Ranking_System # (kind: enumerated) has grades that further subdivide by degree # (IBJJF Art. 4.1.2 - white through black belt each carry numbered # degrees/stripes), so a Rank can legitimately carry both # (rankGrade: bjj_black, rankOrdinal: 6). A prior revision of this # shape forbade both being set at once, assuming every enumerated # system was flat (true for menkyo/shogo/USSC, false for BJJ) - see # ontology/martialis.ttl's 2.1.0 skos:historyNote. # Note: grantedBy is optional (often unknown) # HonorAssertion ################################################################# shapes:HonorAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "HonorAssertion"@en ; rdfs:comment "Honor claims: subject required; optional granting organization (object) or granting person (grantedBy)."@en ; sh:targetClass martialis:HonorAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "HonorAssertion subject must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:class martialis:Organization ; sh:message "HonorAssertion object must be an Organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:grantedBy ; sh:class martialis:Person ; sh:message "Honor granted_by references non-existent person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:assertedOrg ; sh:class martialis:Organization ; sh:message "Honor assertion references non-existent organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:honorType ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "honor_type must be a honor-type concept." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:honorGrade ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "honor_grade must be a shogo concept (the only ordered honorific scheme currently defined)." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:atEvent ; sh:class martialis:Event ; sh:message "Honor event_ids reference non-existent event." ] . # Note: object (granting organization) is optional for honors without granting entity # DocumentTransmissionAssertion ################################################################# shapes:DocumentTransmissionAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "DocumentTransmissionAssertion"@en ; rdfs:comment "Document-based learning claims must reference a transmissionDocument Source."@en ; sh:targetClass martialis:DocumentTransmissionAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "DocumentTransmissionAssertion subject must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:transmissionDocument ; sh:minCount 1 ; sh:class martialis:Source ; sh:message "learned_from document_id must reference an existing Source." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:assertedStyle ; sh:class martialis:Style ; sh:message "Document transmission references non-existent style." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:validDuring ; sh:maxCount 1 ; sh:class martialis:TimeInterval ; sh:message "Document transmission period must be a TimeInterval." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:atEvent ; sh:class martialis:Event ; sh:message "Document transmission event_ids reference non-existent event." ] . # IdentityAssertion ################################################################# shapes:IdentityAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "IdentityAssertion"@en ; rdfs:comment "Candidate-identity claims: subject and object must be distinct Persons."@en ; sh:targetClass martialis:IdentityAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "IdentityAssertion subject must be a Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "possibly_same_as must reference an existing Person." ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "identity not self-referential" ; sh:message "possibly_same_as must not reference the person itself." ; sh:select """ SELECT $this WHERE { $this ?p . $this ?p . } """ ] . # StyleRelationAssertion ################################################################# shapes:StyleRelationAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "StyleRelationAssertion"@en ; rdfs:comment "Style-to-style claims: closed styleRelationType vocabulary; subject and object must be distinct styles."@en ; sh:targetClass martialis:StyleRelationAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:class martialis:Style ; sh:message "StyleRelationAssertion subject must be a Style." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:minCount 1 ; sh:class martialis:Style ; sh:message "Style relation references non-existent style." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:styleRelationType ; sh:minCount 1 ; # Since the SKOS vocabulary migration (2.0.0), values are concept # IRIs - checking scheme membership here is now referential # integrity JSON Schema structurally cannot express (it only sees # the enum string, not the IRI's skos:inScheme). See "Enum # membership: concept-scheme checks" in this file's header. sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "StyleRelationAssertion must have a relation type." ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "style relation not self-referential" ; sh:message "related_styles must not reference the style itself." ; sh:select """ SELECT $this WHERE { $this ?s . $this ?s . } """ ] . # AffiliationAssertion ################################################################# shapes:AffiliationAssertionShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "AffiliationAssertion"@en ; rdfs:comment "Typed affiliation claims: closed relation vocabulary; genealogy values are Organization-only."@en ; sh:targetClass martialis:AffiliationAssertion ; sh:property [ sh:severity sh:Violation ; sh:path martialis:subject ; sh:minCount 1 ; sh:or ( [ sh:class martialis:Place ] [ sh:class martialis:Organization ] ) ; sh:message "AffiliationAssertion subject must be a Place or an Organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:minCount 1 ; sh:class martialis:Organization ; sh:message "AffiliationAssertion must reference an existing Organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:affiliationRelation ; sh:minCount 1 ; sh:maxCount 1 ; # See the concept-scheme note on StyleRelationAssertionShape above. sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "AffiliationAssertion must have exactly one relation." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:validDuring ; sh:maxCount 1 ; sh:class martialis:TimeInterval ; sh:message "Affiliation period must be a TimeInterval." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:atEvent ; sh:class martialis:Event ; sh:message "Affiliation event_ids reference non-existent event." ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "affiliation not self-referential" ; sh:message "Affiliation must not reference the entity itself." ; sh:select """ SELECT $this WHERE { $this ?o . $this ?o . } """ ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "genealogy organization-only" ; sh:message "Genealogy affiliation relations (derived_from, split_from, merged_from, replaces, renamed_from) are Organization-only - a Place cannot claim institutional lineage." ; sh:select """ PREFIX martialis: PREFIX affrel: SELECT $this WHERE { $this martialis:affiliationRelation ?rel ; martialis:subject ?s . FILTER (?rel IN (affrel:derived_from, affrel:split_from, affrel:merged_from, affrel:replaces, affrel:renamed_from)) FILTER NOT EXISTS { ?s a martialis:Organization } } """ ] . # InstructorRole ################################################################# shapes:InstructorRoleShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "InstructorRole"@en ; rdfs:comment "Venue-staff records: instructor must be a Person; capacity, style, and period well-formed."@en ; sh:targetClass martialis:InstructorRole ; sh:property [ sh:severity sh:Violation ; sh:path martialis:instructor ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "Place instructors[].person_id must reference an existing Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:instructorPeriod ; sh:maxCount 1 ; sh:class martialis:TimeInterval ; sh:message "Instructor period must be a TimeInterval." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:instructorCapacity ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "InstructorRole must have at most one capacity." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:instructorStyle ; sh:maxCount 1 ; sh:class martialis:Style ; sh:message "instructors[].style_id must reference an existing Style." ] . # OfficerRole ################################################################# shapes:OfficerRoleShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "OfficerRole"@en ; rdfs:comment "Organization leadership records: officer must be a Person; title free text; period required."@en ; sh:targetClass martialis:OfficerRole ; sh:property [ sh:severity sh:Violation ; sh:path martialis:officer ; sh:minCount 1 ; sh:class martialis:Person ; sh:message "Organization officers[].person_id must reference an existing Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:officerTitle ; sh:minCount 1 ; sh:maxCount 1 ; sh:message "Officer role must have a title." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:officerPeriod ; sh:maxCount 1 ; sh:class martialis:TimeInterval ; sh:message "Officer period must be a TimeInterval." ] . # Place ################################################################# shapes:PlaceShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Place"@en ; rdfs:comment "Venue constraints: required name, placeType vocabulary, parent/affiliation/staff references."@en ; sh:targetClass martialis:Place ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:message "A Place must have at least one name." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:placeType ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "NB: placeType 'school' means an education institution (secondary/high school); a martial arts school/ryuha is an Organization with orgType 'school'." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:parentPlace ; sh:class martialis:Place ; sh:message "parent_place_id references non-existent place." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasAffiliationAssertion ; sh:class martialis:AffiliationAssertion ; sh:message "Place affiliations must be AffiliationAssertion nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasInstructorRole ; sh:class martialis:InstructorRole ; sh:message "Place instructors must be InstructorRole nodes." ] . # Organization ################################################################# shapes:OrganizationShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Organization"@en ; rdfs:comment "Organization constraints: required name, orgType vocabulary, affiliations, officers, founded/dissolved events."@en ; sh:targetClass martialis:Organization ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:message "An Organization must have at least one name." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:orgType ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "NB: orgType 'school' means a martial arts school/ryuha; an education institution is a Place with placeType 'school'." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasAffiliationAssertion ; sh:class martialis:AffiliationAssertion ; sh:message "Organization affiliations must be AffiliationAssertion nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:parentOrg ; sh:class martialis:Organization ; sh:message "parentOrg (materialized from a branch_of affiliation) references non-existent organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasOfficerRole ; sh:class martialis:OfficerRole ; sh:message "Organization officers must be OfficerRole nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:foundedEvent ; sh:maxCount 1 ; sh:class martialis:LifeEvent ; sh:message "founded must be a LifeEvent node." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:dissolvedEvent ; sh:maxCount 1 ; sh:class martialis:LifeEvent ; sh:message "dissolved must be a LifeEvent node." ] . # ChangeEvent (org:ChangeEvent, materialized at export time from # genealogy AffiliationAssertions - see bin/martialis and # ontology/martialis.ttl's "Organizational Change Events" section) ################################################################# shapes:ChangeEventShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "org:ChangeEvent (materialized)"@en ; rdfs:comment "Materialized genealogy events: original/resulting organizations plus changeKind/changeYear."@en ; sh:targetClass org:ChangeEvent ; sh:property [ sh:severity sh:Violation ; sh:path org:originalOrganization ; sh:minCount 1 ; sh:class martialis:Organization ; sh:message "org:originalOrganization must reference an existing Organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path org:resultingOrganization ; sh:minCount 1 ; sh:maxCount 1 ; sh:class martialis:Organization ; sh:message "org:resultingOrganization must reference an existing Organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:changeKind ; sh:minCount 1 ; sh:maxCount 1 ; # changeKind is a materialized copy of the AffiliationAssertion # genealogy relation that produced this ChangeEvent (bin/martialis's # genealogy pass), so it carries the same affiliation-relation # concept IRIs martialis:affiliationRelation does - restricted here # to the genealogy subset via sh:in, since founding-only TIES # values (member_of/branch_of/recognised_by) can never legally # appear on a ChangeEvent. sh:in ( ) ; sh:message "changeKind must be one of: derived_from, split_from, merged_from, replaces, renamed_from." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:changeYear ; sh:maxCount 1 ; sh:datatype xsd:integer ; sh:message "changeYear must be an integer year." ] . # Style ################################################################# shapes:StyleShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Style"@en ; rdfs:comment "Style constraints: required name, styleType/styleForm vocabularies, rank system, style relations."@en ; sh:targetClass martialis:Style ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:message "A Style must have at least one name." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:styleType ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "classification must be a single value (traditional, modern, or other)." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:styleForm ; # No shape existed here before the SKOS vocabulary migration - # values were unenforced free text. Added alongside the concept-IRI # migration since a class/scheme check is now cheap to state. sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "form values must be hybrid and/or sport." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:rankSystem ; sh:maxCount 1 ; sh:class martialis:RankingSystem ; sh:message "rank_system_id must reference an existing RankingSystem entity." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:confersRank ; sh:maxCount 1 ; sh:datatype xsd:boolean ; sh:message "confers_rank must be a boolean." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:foundedEvent ; sh:maxCount 1 ; sh:class martialis:LifeEvent ; sh:message "founded must be a LifeEvent node." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:dissolvedEvent ; sh:maxCount 1 ; sh:class martialis:LifeEvent ; sh:message "dissolved must be a LifeEvent node." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasStyleRelationAssertion ; sh:class martialis:StyleRelationAssertion ; sh:message "related_styles must be expressed via StyleRelationAssertion." ] . # LifeEvent (birth/death on Person; founded/dissolved on Style, # Organization, Place) ################################################################# shapes:LifeEventShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "LifeEvent"@en ; rdfs:comment "Birth/death/founding/dissolution events: founder, evidence, and dissolution manner/reason consistency."@en ; sh:targetClass martialis:LifeEvent ; sh:property [ sh:severity sh:Violation ; sh:path martialis:founder ; sh:class martialis:Person ; sh:message "founder_ids references a non-existent Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasEvidence ; sh:class martialis:SourceAssertionRelation ; sh:message "LifeEvent sources must be SourceAssertionRelation nodes." ] ; # dissolutionManner/dissolutionReason are Organization-only in the # JSON Schema (org.schema.json's dissolved object), but LifeEvent is # shared with birth/death/founded, which simply never populate them - # constraining shape here (rather than a separate Organization-only # dissolved shape) mirrors how founder_ids/place_id/geo are already # handled on this same shape. sh:property [ sh:severity sh:Violation ; sh:path martialis:dissolutionManner ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "Organization must have at most one dissolution manner." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:dissolutionReason ; sh:maxCount 1 ; sh:datatype xsd:string ; sh:message "dissolved.reason must be a string." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:atEvent ; sh:class martialis:Event ; sh:message "LifeEvent event_ids reference non-existent event." ] . # Event ################################################################# shapes:EventShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Event"@en ; rdfs:comment "Notable events: required name, eventType vocabulary, series containment, participation records."@en ; sh:targetClass martialis:Event ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:message "An Event must have a name." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:eventType ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "Event must have at most one classification." ] ; sh:property [ sh:severity sh:Violation ; sh:path schema:organizer ; sh:class martialis:Organization ; sh:message "Event org_ids reference non-existent organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path schema:about ; sh:class martialis:Style ; sh:message "Event style_ids reference non-existent style." ] ; sh:property [ sh:severity sh:Violation ; sh:path schema:superEvent ; sh:class martialis:Event ; sh:message "Event part_of_id references non-existent event." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasParticipation ; sh:class martialis:Participation ; sh:message "Event participants must be Participation nodes." ] . # LifeEvent/Event location integrity ################################################################# # place_id and geo share the schema:location predicate (a reference to # a Place entity vs. an inline, untyped GeoRef blank node): filter to # IRI-valued location references only, so the untyped GeoRef siblings # aren't flagged. Both target classes need the identical check # (LifeEvent's birth/death/founded, Event's own location) - one shared # shape rather than a byte-identical copy on each. shapes:LocationReferencesPlaceShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Location references Place"@en ; rdfs:comment "IRI-valued schema:location on LifeEvent and Event must resolve to a Place entity; inline GeoRef blank nodes are exempt."@en ; sh:targetClass martialis:LifeEvent, martialis:Event ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "place missing reference" ; sh:message "place_id references a non-existent Place." ; sh:select """ PREFIX martialis: PREFIX schema: SELECT $this ?place WHERE { $this schema:location ?place . FILTER(isIRI(?place)) FILTER NOT EXISTS { ?place a martialis:Place } } """ ] . # Country value integrity ################################################################# # schema:addressCountry (nationality codes, inline geo) holds the EU # Publications Office country authority's own concept IRIs, minted by # concatenation at export time (bin/martialis's resolve_country_terms, # EU_COUNTRY_IRI_BASE + code from ontology/country-codes.txt), or the # local exception concept (XRY) from the country scheme in # vocabularies.ttl. Martialis does not vendor the EU authority's RDF # (see docs/VALIDATION.md), so this shape cannot check concept # membership (sh:class skos:Concept) against a local copy - it checks # IRI *shape* instead: one of the two known country-IRI namespaces # followed by a 3-letter code. A bare string or a value from any other # namespace fails. The rdf step already refuses to emit an unknown # code (against ontology/country-codes.txt), so this shape is the # backstop for anything that slips past it. shapes:AddressCountryShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "address country value"@en ; rdfs:comment "Every schema:addressCountry value must be a country IRI: an EU Publications Office country authority IRI or a local exception (XRY) - never a bare code string."@en ; sh:targetSubjectsOf schema:addressCountry ; sh:property [ sh:severity sh:Violation ; sh:path schema:addressCountry ; sh:nodeKind sh:IRI ; sh:pattern "^(http://publications\\.europa\\.eu/resource/authority/country/[A-Z]{3}|https://w3id\\.org/martialis/vocab/country/[A-Z]{3})$" ; sh:message "addressCountry must be an EU authority country IRI or a local exception IRI (XRY), not a bare code string." ] . # Participation ################################################################# shapes:ParticipationShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Participation"@en ; rdfs:comment "Event participation records: object must be a Person or an Organization, role an event-role concept, evidence SourceAssertionRelation nodes."@en ; sh:targetClass martialis:Participation ; sh:property [ sh:severity sh:Violation ; sh:path martialis:object ; sh:minCount 1 ; sh:or ( [ sh:class martialis:Person ] [ sh:class martialis:Organization ] ) ; sh:message "Event participant (person_id/org_id) must be an existing Person or Organization." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:participationRole ; sh:minCount 1 ; # Event participation roles are their own vocabulary (event-role), # split off martialis:role in the 2.1.0 quality pass - teacher-role # (student/uchi-deshi/coach/...) is for TeacherAssertions only. sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "participants[].role must be an event-role concept (competitor, official, organizer, demonstrator, guest, instructor, student, other)." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasEvidence ; sh:class martialis:SourceAssertionRelation ; sh:message "Participation sources must be SourceAssertionRelation nodes." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:participationStyle ; sh:class martialis:Style ; sh:message "participants[].style_ids must reference an existing Style." ] . # TimeInterval ################################################################# shapes:TimeIntervalShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "TimeInterval"@en ; rdfs:comment "startYear must not be later than endYear."@en ; sh:targetClass martialis:TimeInterval ; sh:property [ sh:severity sh:Violation ; sh:path martialis:startYear ; sh:datatype xsd:integer ; sh:message "startYear must be an integer year." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:endYear ; sh:datatype xsd:integer ; sh:message "endYear must be an integer year." ] ; sh:sparql [ sh:name "interval end before start" ; sh:message "endYear must not be earlier than startYear." ; sh:select """ PREFIX xsd: SELECT $this WHERE { $this ?s . $this ?e . FILTER (xsd:integer(STR(?e)) < xsd:integer(STR(?s))) } """ ] . # SourceAssertionRelation (Evidence) ################################################################# shapes:SourceAssertionRelationShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "SourceAssertionRelation (evidence)"@en ; rdfs:comment "Evidence records: basedOnSource required; supports, weight, locator type, and quotation consistency."@en ; sh:targetClass martialis:SourceAssertionRelation ; sh:property [ sh:severity sh:Violation ; sh:path martialis:basedOnSource ; sh:class martialis:Source ; sh:message "Evidence source_id must reference an existing Source." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:supports ; sh:minCount 1 ; sh:datatype xsd:boolean ; sh:message "Evidence must state whether it supports or contradicts the claim." ] ; # sourceRefType and evidenceWeight are concept IRIs since the SKOS # vocabulary migration (2.0.0) - sh:class/sh:node replace their old # sh:datatype xsd:string checks. sourceLocatorLabel stays a plain # string below: it aligns to CSL's citation-locator vocabulary, an # external standard, so under the SKOS plan's governing rule # ("concept IRIs for vocabularies Martialis defines, literals for # tokens defined by an external standard") it was deliberately left # out of the migration - see ontology/vocabularies/source-locator-label.ttl. sh:property [ sh:severity sh:Violation ; sh:path martialis:sourceRefType ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "SourceAssertionRelation must have a sourceRefType." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:evidenceWeight ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "SourceAssertionRelation must have an evidenceWeight." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:sourceLocatorLabel ; sh:datatype xsd:string ; sh:message "sourceLocatorLabel must be a string." ] ; sh:property [ sh:severity sh:Violation ; sh:path schema:archivedAt ; sh:nodeKind sh:IRI ; sh:maxCount 1 ; sh:message "archive_uri must be a valid IRI (Wayback Machine URL)." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:coversProperty ; sh:nodeKind sh:IRI ; sh:in ( schema:name schema:alternateName schema:nationality schema:location schema:address schema:description schema:image schema:logo schema:sameAs schema:url schema:startDate schema:organizer schema:about martialis:researchNote martialis:birthEvent martialis:deathEvent martialis:foundedEvent martialis:dissolvedEvent martialis:hasTeacherAssertion martialis:hasContextualTrainingAssertion martialis:hasRankAssertion martialis:hasHonorAssertion martialis:hasDocumentTransmissionAssertion martialis:hasIdentityAssertion martialis:hasStyleRelationAssertion martialis:hasAffiliationAssertion martialis:hasParticipation martialis:rankSystem martialis:styleType martialis:styleForm martialis:orgType martialis:placeType martialis:eventType martialis:sourceType martialis:parentPlace martialis:hasInstructorRole martialis:hasOfficerRole martialis:rankingSystemKind martialis:gradeScheme martialis:minOrdinal martialis:maxOrdinal martialis:confersRank martialis:derivedFromSystem ) ; sh:message "covers must name a real entity property (a closed set of predicate IRIs); an unrecognized token would otherwise resolve to a junk IRI under the @base entity namespace instead of a real predicate." ] . # Confidence ################################################################# shapes:ConfidenceShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Confidence"@en ; rdfs:comment "Confidence assessments: level within 0-3, assessedBy an agent IRI."@en ; sh:targetClass martialis:Confidence ; sh:property [ sh:severity sh:Violation ; sh:path martialis:confidenceLevel ; sh:maxCount 1 ; # Enum membership not checked here - martialis.schema.json's # Confidence.level already guarantees 0/1/2/3 (Disputed/ # Uncertain/Probable/Confirmed). sh:message "Confidence must have at most one level." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:assessedBy ; sh:nodeKind sh:IRI ; sh:pattern "^https://martial\\.is/agent/" ; sh:message "assessedBy must be an agent IRI under https://martial.is/agent/ (see martialis:Agent)." ] . # Source ################################################################# shapes:SourceShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Source"@en ; rdfs:comment "Bibliographic constraints: CSL sourceType vocabulary, genre, peer-review and self-publication values, containment vs edition structure, identifiers."@en ; sh:targetClass martialis:Source ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:message "A Source must have a title." ] ; # sourceType and peerReview stay unchecked here deliberately, not # because JSON Schema already covers them (that reasoning no longer # applies generally post-migration - see the top-of-file policy note) # but because both are external-standard alignments under the SKOS # plan's governing rule and were excluded from the concept-IRI # migration: sourceType is the CSL-JSON type vocabulary verbatim # (ontology/vocabularies/source-classification.ttl, no per-value # concepts), and peerReview is a DRIVER/OpenAIRE value alignment # (ontology/vocabularies/peer-review.ttl) that has carried this same # "value alignment only, property stays a string" design since before # this migration. sourceGenre, below, IS migrated and IS checked. sh:property [ sh:severity sh:Violation ; sh:path martialis:sourceType ; sh:maxCount 1 ; sh:message "A Source must have at most one classification." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:sourceGenre ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "A Source must have at most one genre." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:peerReview ; sh:maxCount 1 ; sh:message "A Source must have at most one peer_review value." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:selfPublished ; sh:maxCount 1 ; sh:datatype xsd:boolean ; sh:message "self_published must be a boolean." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:partOf ; sh:maxCount 1 ; sh:class martialis:Source ; sh:message "part_of_id must reference an existing Source." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:editionOf ; sh:maxCount 1 ; sh:node shapes:EditionOfShape ; sh:message "edition_of must be a valid edition/translation/reprint/facsimile relation." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasContributor ; sh:node shapes:ContributorShape ; sh:message "Each contributor must have a valid role, and person_id/org_id (if present) must reference an existing entity." ] ; sh:property [ sh:path martialis:volume ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:issue ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:page ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:edition ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:chapterNumber ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:partNumber ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:section ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:archive ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:archiveCollection ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:archiveLocation ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:callNumber ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:accessed ; sh:maxCount 1 ] ; sh:property [ sh:path martialis:archiveDate ; sh:maxCount 1 ] ; sh:property [ sh:path schema:datePublished ; sh:maxCount 1 ] ; sh:property [ sh:severity sh:Violation ; sh:path schema:publisher ; sh:maxCount 1 ; sh:node shapes:PublisherShape ; sh:message "publisher must have a name, and org_id (if present) must reference an existing Organization." ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "source part-of cycle" ; sh:message "part_of must not form a cycle - a Source cannot, directly or transitively, be part of itself." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:partOf+ $this . } """ ] . shapes:EditionOfShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Edition relation"@en ; rdfs:comment "Edition nodes: editionOfSource required, closed editionRelationType vocabulary."@en ; sh:property [ sh:severity sh:Violation ; sh:path martialis:editionOfSource ; sh:minCount 1 ; sh:maxCount 1 ; sh:class martialis:Source ; sh:message "edition_of.source_id must reference an existing Source." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:editionRelationType ; sh:minCount 1 ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "edition_of must have exactly one relation." ] . shapes:PublisherShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Publisher"@en ; rdfs:comment "Publisher nodes: a name or a publisherOrg reference."@en ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:maxCount 1 ; sh:message "publisher.name is required." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:publisherOrg ; sh:maxCount 1 ; sh:class martialis:Organization ; sh:message "publisher.org_id must reference an existing Organization." ] . shapes:ContributorShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Contributor"@en ; rdfs:comment "Contributor nodes: CSL contributor role; person/org cross-references."@en ; sh:property [ sh:severity sh:Violation ; sh:path martialis:contributorRole ; sh:minCount 1 ; sh:maxCount 1 ; # Enum membership not checked here - source.schema.json's # Contributor.role already guarantees the same 26 CSL values. sh:message "A contributor must have exactly one role." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:contributorPerson ; sh:maxCount 1 ; sh:class martialis:Person ; sh:message "contributor person_id must reference an existing Person." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:contributorOrg ; sh:maxCount 1 ; sh:class martialis:Organization ; sh:message "contributor org_id must reference an existing Organization." ] . # RankSystem, and the two properties left unchecked at the end of the # SKOS vocabulary migration (rankSystemType, dateCalendar, linkType - # added here as a deliberate final pass, not overlooked the first time). ################################################################# shapes:RankingSystemShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "RankingSystem"@en ; rdfs:comment "RankingSystem is a first-class, lightly-sourced entity (database/ranking-systems/), the same shape as Style/Organization/Place - name and kind required; grade_scheme required only when kind is 'enumerated'."@en ; sh:targetClass martialis:RankingSystem ; sh:property [ sh:severity sh:Violation ; sh:path schema:name ; sh:minCount 1 ; sh:message "A RankingSystem must have at least one name." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:foundedEvent ; sh:maxCount 1 ; sh:class martialis:LifeEvent ; sh:message "instituted must be a LifeEvent node." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:hasEvidence ; sh:class martialis:SourceAssertionRelation ; sh:message "sources must be expressed as evidence records." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:derivedFromSystem ; sh:maxCount 1 ; sh:class martialis:RankingSystem ; sh:message "derived_from_id must reference an existing RankingSystem entity." ] ; sh:property [ sh:severity sh:Violation ; sh:path martialis:rankingSystemKind ; sh:minCount 1 ; sh:maxCount 1 ; sh:in ( "numbered" "enumerated" ) ; sh:message "kind must be 'numbered' or 'enumerated'." ] ; sh:or ( [ sh:property [ sh:path martialis:rankingSystemKind ; sh:hasValue "numbered" ] ] [ sh:property [ sh:path martialis:gradeScheme ; sh:minCount 1 ] ] ) ; sh:message "An 'enumerated' RankingSystem must declare its grade_scheme." . # dateCalendar and linkType cut across several unrelated node types # (dateCalendar: LifeEvent, RankAssertion, HonorAssertion, Source; # linkType: an arbitrary external resource, the object of schema:sameAs, # never a Martialis-typed node at all - see schemas/context.jsonld's # links term, "uri": "@id") - sh:targetClass cannot express either, so # both shapes target every SUBJECT of the property instead, wherever in # the graph it occurs. shapes:DateCalendarShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "date calendar usage"@en ; rdfs:comment "Any node carrying martialis:dateCalendar (LifeEvent, RankAssertion, HonorAssertion, Source): concept-scheme membership."@en ; sh:targetSubjectsOf martialis:dateCalendar ; sh:property [ sh:severity sh:Violation ; sh:path martialis:dateCalendar ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "date_calendar must be a date-calendar concept." ] . shapes:LinkTypeShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "link type usage"@en ; rdfs:comment "Any node carrying martialis:linkType (the external resource a links[] entry names, schema:sameAs's object - never a Martialis-typed node): concept-scheme membership."@en ; sh:targetSubjectsOf martialis:linkType ; sh:property [ sh:severity sh:Violation ; sh:path martialis:linkType ; sh:maxCount 1 ; sh:class skos:Concept ; sh:node [ sh:property [ sh:path skos:inScheme ; sh:hasValue ] ] ; sh:message "links[].type must be a link-type concept." ] . # Locator range ordering ################################################################# # The FORMAT of a locator is JSON Schema's job (SourceRef's allOf # requires a numeric-label locator to be a number, a dotted number, or an # N-M range, and a timestamp label to look like a clock time). What a # regex cannot express is arithmetic: "42-17" is well-formed and wrong. # # Targeted by sh:targetSubjectsOf rather than a class, because a locator # can sit on any evidence-shaped node - including photo/video attribution, # which is deliberately never typed SourceAssertionRelation. # # Dotted forms are excluded on both sides on purpose: "1.2-1.10" is not # numerically comparable as a whole, so ordering is not asserted for it. shapes:LocatorRangeShape a sh:NodeShape ; sh:severity sh:Violation ; rdfs:isDefinedBy ; rdfs:label "Locator range ordering"@en ; rdfs:comment "A numeric-label locator written as a range must ascend."@en ; sh:targetSubjectsOf martialis:sourceLocation ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "locator range not ascending" ; sh:message "A numeric locator range must ascend - this one runs backwards (e.g. page 42-17)." ; sh:select """ PREFIX martialis: PREFIX xsd: SELECT $this WHERE { $this martialis:sourceLocation ?loc ; martialis:sourceLocatorLabel ?label . FILTER(?label IN ("page", "volume", "issue", "line", "paragraph", "folio", "note", "table", "figure", "verse", "section", "equation", "opus", "act", "column", "elocation", "supplement")) FILTER(REGEX(STR(?loc), "^\\\\s*[0-9]+\\\\s*-\\\\s*[0-9]+\\\\s*$")) BIND(xsd:integer(REPLACE(STR(?loc), "^\\\\s*([0-9]+)\\\\s*-.*$", "$1")) AS ?from) BIND(xsd:integer(REPLACE(STR(?loc), "^.*-\\\\s*([0-9]+)\\\\s*$", "$1")) AS ?to) FILTER(?from > ?to) } """ ] . ### Extra SHACL ### ### Quality and completeness validation for the Martialis database ### ### Frederico Munoz , Martialis Project ### ### Licence: CC0-1.0 (public domain dedication) ### Attribution requested but not required: cite "Martialis — https://martial.is" ### ### These shapes generate warnings for incomplete data that should ### be investigated and improved, but don't block publication. ### ### Since ontology 2.3.0 the JSON Schemas REQUIRE at least one source ### (minItems 1) on every assertion-shaped node: rank, honor, ### learned_from, possibly_same_as, related_styles, affiliations, ### instructors, officers, event participations, and every teacher ### interaction whose teacher entry carries a real teacher_id (the ### unknown-teacher placeholder, teacher_id: null, is dropped at RDF ### export and stays exempt). The "has no sources" Warning checks below ### can therefore no longer fire on schema-valid data; they are ### retained as a belt-and-braces check for hand-built RDF that never ### passed through the schemas. ### ### Per-source-ref field completeness (type, weight, quotation, ### location, citation) is intentionally NOT checked here: it is ### aggregate information, reported as percentages by the provenance ### and dashboard subcommands of bin/martialis, where it does not ### drown the per-item findings below. @prefix sh: . @prefix rdf: . @prefix xsd: . @prefix martialis: . @prefix shapes: . @prefix schema: . @prefix prov: . @prefix rdfs: . @prefix owl: . @prefix dcterms: . @prefix vann: . # Shapes graph self-description ################################################################# # Declared for documentation generators (pyLODE valpub) and for # self-description; annotation triples here do not affect validation. a owl:Ontology , sh:ShapesGraph ; dcterms:title "Martialis Quality and Completeness Constraints"@en ; dcterms:description "Quality and completeness validation for the Martialis database. These shapes generate warnings for incomplete data that should be investigated and improved, but do not block publication - the structural tier (martialis.shacl.ttl) is the blocking gate. Per-source-ref field completeness is intentionally not checked here: it is aggregate information, reported by the provenance and dashboard subcommands of bin/martialis."@en ; dcterms:license ; dcterms:created "2026-02-28"^^xsd:date ; dcterms:modified "2026-09-07"^^xsd:date ; owl:versionInfo "1.2.0" ; owl:imports ; vann:preferredNamespacePrefix "shapes" ; vann:preferredNamespaceUri ; rdfs:seeAlso . # Person Completeness ################################################################# shapes:PersonCompletenessShape a sh:NodeShape ; sh:severity sh:Warning ; rdfs:isDefinedBy ; rdfs:label "Person completeness"@en ; rdfs:comment "Warns on Persons missing a birth date or description; birthEvent existence itself is schema-guaranteed."@en ; sh:targetClass martialis:Person ; # Either a named teacher or a contextual training claim satisfies # this check - a Person with only teacher_id: null entries (no # identified teacher, but a sourced place/event/period claim) is not # missing lineage information the way a Person with neither has. sh:or ( [ sh:property [ sh:path martialis:hasTeacherAssertion ; sh:minCount 1 ] ] [ sh:property [ sh:path martialis:hasContextualTrainingAssertion ; sh:minCount 1 ] ] ) ; sh:message "Quality check: Person has no teachers or contextual training - needs investigation." ; sh:property [ sh:path martialis:hasRankAssertion ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: Person has no rank information - consider adding if available." ] ; sh:property [ sh:path (martialis:birthEvent schema:birthDate) ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: Person has no birth date - consider adding if available." ] . # Note: birthEvent's mere existence is not checked here (nor was an # analogous "has TeacherAssertion.object" check kept) - JSON Schema's # own `required` already guarantees the node/property exists wherever # the schema demands it, so a SHACL minCount on the same predicate can # never fire and would be dead weight. This shape instead follows the # sequence path through to schema:birthDate, the field that actually # varies (a Person can have `birth: {date: null}` - a birth node with # no date - which the old node-presence check read as "complete"). # Signal, not a violation: co-teaching, posthumous transmission via # students, and simple recording errors can all produce this pattern, # so it needs a human look rather than blocking. A standalone shape # because sh:severity on the shape is what pyshacl actually honors for # a SPARQLConstraint - a severity declared inside sh:sparql itself is # silently ignored and falls back to sh:Violation (confirmed # empirically), and a shared NodeShape would wrongly apply this # severity to its sibling property constraints too. shapes:TeacherAgeQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Teacher age plausibility"@en ; rdfs:comment "Signals teacher-student interactions whose dates imply an implausible age difference."@en ; sh:targetClass martialis:TeacherAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "teacher born after student" ; sh:message "Quality check: the teacher was born after the student - worth double-checking." ; sh:select """ PREFIX martialis: PREFIX xsd: SELECT $this WHERE { $this martialis:subject ?student ; martialis:object ?teacher . ?student martialis:birthEvent/martialis:latestYear ?studentLatest . ?teacher martialis:birthEvent/martialis:earliestYear ?teacherEarliest . FILTER(xsd:integer(STR(?teacherEarliest)) > xsd:integer(STR(?studentLatest))) } """ ] . # TeacherInteraction Completeness ################################################################# shapes:TeacherInteractionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "TeacherInteraction completeness"@en ; rdfs:comment "Warns when interactions lack a place or a period."@en ; sh:targetClass martialis:TeacherInteraction ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: TeacherInteraction has no sources - needs investigation and sourcing." ] ; sh:property [ sh:path martialis:interactionPeriod ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: TeacherInteraction has no time period - consider adding if available." ] . # See the note on TeacherAgeQualityShape above for why this is a # separate shape rather than a third constraint on # TeacherInteractionCompletenessShape. shapes:InteractionPeriodQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Interaction period plausibility"@en ; rdfs:comment "Signals interaction periods inconsistent with the student's dates."@en ; sh:targetClass martialis:TeacherInteraction ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "interaction outside lifespan" ; sh:message "Quality check: interaction period falls outside the student's or teacher's lifespan - worth double-checking." ; # NB: the FILTERs must sit at TOP level, outside the UNION # groups - rdflib (this pipeline's SPARQL engine) fails to # match a FILTER placed inside a group when it references a # variable bound in the enclosing scope, which silently made # this constraint fire on nothing (caught by the # tests/shacl-fixtures/ regression gate). sh:select """ PREFIX martialis: PREFIX xsd: SELECT $this WHERE { ?assertion martialis:hasInteraction $this ; martialis:subject ?student ; martialis:object ?teacher . $this martialis:interactionPeriod ?period . ?period martialis:startYear ?start . { { ?student martialis:birthEvent/martialis:earliestYear ?bound . BIND("lower" AS ?kind) } UNION { ?student martialis:deathEvent/martialis:latestYear ?bound . BIND("upper" AS ?kind) } UNION { ?teacher martialis:birthEvent/martialis:earliestYear ?bound . BIND("lower" AS ?kind) } UNION { ?teacher martialis:deathEvent/martialis:latestYear ?bound . BIND("upper" AS ?kind) } } FILTER ( (?kind = "lower" && xsd:integer(STR(?start)) < xsd:integer(STR(?bound))) || (?kind = "upper" && xsd:integer(STR(?start)) > xsd:integer(STR(?bound))) ) } """ ] . # A teacher relationship is one of the most load-bearing claims in the # database (it materializes into the lineage tree); "has a source" isn't # enough on its own if none of those sources actually quote the claim. # Separate shape for the same severity-declaration reason noted on # TeacherAgeQualityShape above. shapes:TeacherInteractionQuotationQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Teacher evidence quotation"@en ; rdfs:comment "Warns when a teacher claim's sources carry no quotation - sourced only in paraphrase."@en ; sh:targetClass martialis:TeacherInteraction ; sh:severity sh:Warning ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "interaction missing quotation" ; sh:message "Quality check: TeacherInteraction has sources, but none carry a quotation - a direct quote substantiating the relationship is expected." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:hasEvidence ?e . FILTER NOT EXISTS { $this martialis:hasEvidence/martialis:quotation ?q } } """ ] . # Softer than the quotation check above (Info, not Warning): citing # Wikipedia isn't wrong, but a teacher relationship resting SOLELY on # Wikipedia risks circularity - community lineage lore often circulates # back through Wikipedia itself. Conservatively requires every evidence # entry to have a resolvable wikipedia.org URL (its own inline uri, or # its basedOnSource's url) before firing - an entry with no URL at all # (e.g. an offline book citation) correctly suppresses the flag rather # than being silently ignored, avoiding false positives on entries that # already have real non-Wikipedia corroboration. shapes:TeacherInteractionWikipediaOnlyQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Wikipedia-only teacher evidence"@en ; rdfs:comment "Info: teacher claims whose every source is Wikipedia - circularity risk."@en ; sh:targetClass martialis:TeacherInteraction ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "interaction wikipedia-only sources" ; sh:message "Quality check: every source on this TeacherInteraction is a Wikipedia article - consider adding a non-Wikipedia corroborating source to avoid circularity." ; sh:select """ PREFIX martialis: PREFIX schema: SELECT $this WHERE { $this martialis:hasEvidence ?e . FILTER NOT EXISTS { $this martialis:hasEvidence ?e2 . FILTER NOT EXISTS { { ?e2 martialis:basedOnSource/schema:url ?u2 } UNION { ?e2 schema:url ?u2 } FILTER(CONTAINS(STR(?u2), "wikipedia.org")) } } } """ ] . # ContextualTrainingAssertion Completeness ################################################################# # The self-improving signal this class exists for: a contextual claim # names a venue and a window but not a teacher, and a Place's # InstructorRole roster is exactly the place that a candidate might # already be recorded, or where it is worth investigating. Only fires # for place-anchored interactions - an event-anchored claim's # candidates are the event's own instructor participations, not this # join, and are a query for the consuming application rather than a # curation check. shapes:ContextualTrainingAssertionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Contextual training candidate instructors"@en ; rdfs:comment "Info: a contextual training claim's venue has no InstructorRole overlapping its period - no candidate instructors are recorded for this venue and window."@en ; sh:targetClass martialis:ContextualTrainingAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "contextual training no candidate instructors" ; sh:message "Quality check: no InstructorRole on this venue overlaps the claimed period - no candidate instructors are recorded for this venue and window; worth researching." ; sh:select """ PREFIX martialis: PREFIX xsd: SELECT $this WHERE { $this martialis:hasInteraction ?i . ?i martialis:interactionPlace ?place ; martialis:interactionPeriod ?period . ?period martialis:startYear ?start . FILTER NOT EXISTS { ?place martialis:hasInstructorRole ?role . ?role martialis:instructorPeriod ?rperiod . ?rperiod martialis:startYear ?rstart . OPTIONAL { ?rperiod martialis:endYear ?rend } FILTER(xsd:integer(STR(?rstart)) <= xsd:integer(STR(?start)) && (!BOUND(?rend) || xsd:integer(STR(?rend)) >= xsd:integer(STR(?start)))) } } """ ] . # Participation Style Attribution ################################################################# # A multi-style course/seminar's event-level style_ids are an unattributed # set: they cannot say which instructor taught which art. The roster # sources usually do name it ("Judo instruction was provided by Kotani, # Otaki..."), so the per-participation martialis:participationStyle is the # only place that mapping can live, and without it candidate-instructor # matches against a student's contextual-training style_id degrade to # "any instructor at this event, any style". This Info points at exactly # the participations worth annotating - the same self-improving pattern # as ContextualTrainingAssertionCompletenessShape above. shapes:ParticipationStyleAttributionShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Participation style attribution"@en ; rdfs:comment "Info: a multi-style course/seminar whose instructor participations lack style annotations."@en ; sh:targetClass martialis:Event ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "instructor participation missing style" ; sh:message "Quality check: this course/seminar covers multiple styles but an instructor participation carries no style_ids - annotate which art that instructor taught here so candidate matches are attributable." ; sh:select """ PREFIX martialis: PREFIX schema: PREFIX eventclass: PREFIX eventrole: SELECT $this WHERE { $this a martialis:Event ; martialis:eventType ?type ; schema:about ?s1 , ?s2 ; martialis:hasParticipation ?p . ?p martialis:participationRole eventrole:instructor . FILTER (?type IN (eventclass:course, eventclass:seminar)) FILTER (?s1 != ?s2) FILTER NOT EXISTS { ?p martialis:participationStyle ?st } } """ ] . # Document Transmission Author Link ################################################################# # learned_from (DocumentTransmissionAssertion) is reader -> document; the # reader -> author relation the webapp materializes is derived by joining # the document's contributors[].person_id, filtered to the authorship # roles (author, original-author, container-author - contributorRole is # CSL's 26-value vocabulary, and a translator's or interviewer's # person_id does not make them the work's author). A named authorship # contributor without that link is exactly where the derivation stops # short - this Info points at it, the same self-improving pattern as the # two shapes above. shapes:DocumentTransmissionAuthorLinkShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Document transmission author link"@en ; rdfs:comment "Info: a document-transmission claim's document names an authorship-role contributor who is not linked to a Person record."@en ; sh:targetClass martialis:DocumentTransmissionAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "transmission document author unlinked" ; sh:message "Quality check: this learned_from document's author is named but not linked to a Person - add contributors[].person_id when the author is a Person record (this is how the reader-to-author relation materializes)." ; sh:select """ PREFIX martialis: PREFIX schema: SELECT $this WHERE { $this martialis:transmissionDocument ?doc . ?doc martialis:hasContributor ?c . ?c martialis:contributorRole ?role . FILTER (?role IN ("author", "original-author", "container-author")) FILTER EXISTS { { ?c schema:familyName ?f } UNION { ?c schema:givenName ?g } UNION { ?c schema:name ?n } } FILTER NOT EXISTS { ?c martialis:contributorPerson ?p } } """ ] . # Date Precision & Calendar Provenance ################################################################# # Both non-blocking: an editorial nudge, not a data error. See # ontology/martialis.ttl's "Date precision & calendar provenance" # section for what dateOriginal/dateCalendar/earliestYear/latestYear # mean and why they carry no single rdfs:domain. # Uses sh:targetSubjectsOf rather than duplicating this check once per # Date-bearing class (LifeEvent, RankAssertion, HonorAssertion, # foundedEvent/dissolvedEvent, Source): every node that has a # dateCalendar/issued_calendar value at all is in scope, regardless of # which class it belongs to. shapes:DateCalendarCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Date calendar provenance"@en ; rdfs:comment "Info: dateCalendar recorded without the paired dateOriginal."@en ; sh:targetSubjectsOf martialis:dateCalendar ; sh:property [ sh:path martialis:dateOriginal ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: date_calendar is set without date_original - the source's own rendering should usually be recorded alongside the calendar classification." ] . # A crude, English-only heuristic, restricted to RankAssertion/ # HonorAssertion (the only classes where a date sits alongside a # confidence.rationale free-text field): flags prose uncertainty signals # when the date itself carries none of EDTF's own uncertainty markers, as # a nudge to encode the uncertainty structurally (~ ? % an interval or a # Level 2 set) instead of leaving it only in prose. Expected to have # false negatives (it can't catch every way uncertainty gets phrased) and # occasional false positives (rationale text can mention "circa" about # something other than this exact date) - a signal for a curator to look # at, not a fact being asserted. shapes:DatePrecisionQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Date precision"@en ; rdfs:comment "Nudges prose uncertainty in confidence rationales toward structural EDTF uncertainty markers."@en ; sh:targetClass martialis:RankAssertion, martialis:HonorAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "date uncertainty unencoded" ; sh:message "Quality check: confidence.rationale suggests date uncertainty, but the date field itself carries no EDTF uncertainty marker - consider encoding it structurally (~ ? % an interval, or a Level 2 set) instead of only in prose." ; sh:select """ PREFIX martialis: PREFIX schema: SELECT $this WHERE { $this schema:datePublished ?date ; martialis:hasConfidence ?conf . ?conf martialis:confidenceRationale ?rationale . FILTER(!CONTAINS(STR(?date), "~") && !CONTAINS(STR(?date), "?") && !CONTAINS(STR(?date), "%") && !CONTAINS(STR(?date), "/") && !CONTAINS(STR(?date), "[")) FILTER(REGEX(STR(?rationale), "circa|approx|conflicting date|disput|uncertain date|probably (in|around)|before [0-9]{4}", "i")) } """ ] . # RankAssertion Completeness ################################################################# # Teacher evidence lives on TeacherInteraction nodes (checked above); # rank, honor, and style-relation evidence lives on the assertion. # rankSystem, object (style), and hasConfidence are not checked here - # each is already `required` in person.schema.json's Rank/Confidence # definitions, so a SHACL minCount on the same predicate can never # fire. Only hasEvidence (not schema-required - a Rank can validly # have interactions instead) and grantedBy (never required, genuinely # often unknown) carry real curatorial signal. shapes:RankAssertionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Rank assertion completeness"@en ; rdfs:comment "Warns on rank claims without evidence; Info on missing grantedBy."@en ; sh:targetClass martialis:RankAssertion ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: RankAssertion has no sources - needs investigation and sourcing." ] ; sh:property [ sh:path martialis:grantedBy ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: RankAssertion has no grantedBy - consider adding source if available." ] . # See TeacherInteractionQuotationQualityShape - same rationale, applied # to rank grants: a rank claim with a source but no quotation isn't # meaningfully sourced yet. shapes:RankAssertionQuotationQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Rank evidence quotation"@en ; rdfs:comment "Warns when a rank grant's sources carry no quotation."@en ; sh:targetClass martialis:RankAssertion ; sh:severity sh:Warning ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "rank missing quotation" ; sh:message "Quality check: RankAssertion has sources, but none carry a quotation - a direct quote substantiating the grant is expected." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:hasEvidence ?e . FILTER NOT EXISTS { $this martialis:hasEvidence/martialis:quotation ?q } } """ ] . # See TeacherInteractionWikipediaOnlyQualityShape - same rationale and # same conservative "every entry must have a resolvable wikipedia.org # URL" condition, applied to rank grants. shapes:RankAssertionWikipediaOnlyQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Wikipedia-only rank evidence"@en ; rdfs:comment "Info: rank claims sourced solely from Wikipedia."@en ; sh:targetClass martialis:RankAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "rank wikipedia-only sources" ; sh:message "Quality check: every source on this RankAssertion is a Wikipedia article - consider adding a non-Wikipedia corroborating source to avoid circularity." ; sh:select """ PREFIX martialis: PREFIX schema: SELECT $this WHERE { $this martialis:hasEvidence ?e . FILTER NOT EXISTS { $this martialis:hasEvidence ?e2 . FILTER NOT EXISTS { { ?e2 martialis:basedOnSource/schema:url ?u2 } UNION { ?e2 schema:url ?u2 } FILTER(CONTAINS(STR(?u2), "wikipedia.org")) } } } """ ] . # A rank that records neither direction (rankType), number # (rankOrdinal), nor named grade (rankGrade) is a documented-but- # unshaped claim: the schema permits it (a source may only say # "black belt" with no specific degree), so this is a completion # nudge, not a structural defect. shapes:RankDirectionQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Rank direction/grade completeness"@en ; rdfs:comment "Info: ranks with neither rank_type, rank_ordinal nor rank_grade."@en ; sh:targetClass martialis:RankAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "rank missing direction or grade" ; sh:message "Quality check: rank records neither rank_type, rank_ordinal nor rank_grade - consider adding the documented direction or grade if the source states one." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this a martialis:RankAssertion . FILTER NOT EXISTS { $this martialis:rankType ?t } FILTER NOT EXISTS { $this martialis:rankOrdinal ?o } FILTER NOT EXISTS { $this martialis:rankGrade ?g } } """ ] . # HonorAssertion Completeness ################################################################# # hasConfidence is not checked here - already `required` in # person.schema.json's Honor definition, so it can never fire. shapes:HonorAssertionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Honor assertion completeness"@en ; rdfs:comment "Warns on honors without evidence."@en ; sh:targetClass martialis:HonorAssertion ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: HonorAssertion has no sources - needs investigation and sourcing." ] . # StyleRelationAssertion Completeness ################################################################# shapes:StyleRelationAssertionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Style relation completeness"@en ; rdfs:comment "Warns on style relations without evidence."@en ; sh:targetClass martialis:StyleRelationAssertion ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: style relation has no sources - needs investigation and sourcing." ] . # DocumentTransmissionAssertion Completeness ################################################################# shapes:DocumentTransmissionAssertionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Document transmission completeness"@en ; rdfs:comment "Warns on document-learning claims without evidence of the person using the document."@en ; sh:targetClass martialis:DocumentTransmissionAssertion ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: document transmission has no evidence that the person used the document - needs sourcing." ] . # IdentityAssertion Completeness ################################################################# shapes:IdentityAssertionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Identity assertion completeness"@en ; rdfs:comment "Warns on identity candidates without evidence."@en ; sh:targetClass martialis:IdentityAssertion ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: identity candidate has no sources - needs investigation and sourcing." ] . # Place Completeness ################################################################# shapes:PlaceCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Place completeness"@en ; rdfs:comment "Info: venues with no organizational affiliation or founding information."@en ; sh:targetClass martialis:Place ; sh:property [ sh:path martialis:hasAffiliationAssertion ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: Place has no organizational affiliation - consider adding if available." ] ; sh:property [ sh:path martialis:foundedEvent ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: Place has no founding information - consider adding if available." ] . # InstructorRole Completeness ################################################################# # instructorPeriod is not checked here - already `required` in # martialis.schema.json's InstructorRole definition, so it can never fire. shapes:InstructorRoleCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Instructor role completeness"@en ; rdfs:comment "Warns on venue-staff records without sources."@en ; sh:targetClass martialis:InstructorRole ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: instructor role has no sources - needs investigation and sourcing." ] . # OfficerRole Completeness ################################################################# # officerPeriod is not checked here - already `required` in # martialis.schema.json's OfficerRole definition, so it can never fire. shapes:OfficerRoleCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Officer role completeness"@en ; rdfs:comment "Warns on officer records without sources."@en ; sh:targetClass martialis:OfficerRole ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: officer role has no sources - needs investigation and sourcing." ] . # AffiliationAssertion Completeness ################################################################# # Unlike InstructorRole, an AffiliationAssertion had NO completeness # check at all before this: a sourceless, periodless affiliation # (just an org_id and a relation) validated clean at both structural # and quality tiers. Mirrors InstructorRoleCompletenessShape above. shapes:AffiliationAssertionCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Affiliation completeness"@en ; rdfs:comment "Warns on sourceless affiliations; Info on missing period."@en ; sh:targetClass martialis:AffiliationAssertion ; sh:property [ sh:path martialis:hasEvidence ; sh:minCount 1 ; sh:severity sh:Warning ; sh:message "Quality check: affiliation has no sources - needs investigation and sourcing." ] ; sh:property [ sh:path martialis:validDuring ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: affiliation has no period - consider adding if available." ] . # (validDuring, when present, always carries a startYear - Period.start # is schema-required - so the check above already covers a genealogy # affiliation missing the year the ChangeEvent materialization keys on; # no separate "no start year" check is needed.) # Organization Completeness ################################################################# # No OrganizationCompletenessShape existed before this: an Organization # could carry a bare `dissolved: {date}` with no `manner`, and a # genealogy affiliation could point at a predecessor never marked # dissolved (or a dissolved organization could sit unclaimed by any # successor), with no warning at either validation tier. Each check # below is its own NodeShape rather than one shared shape, for the same # reason as TeacherAgeQualityShape above: pyshacl silently ignores a # severity declared inside sh:sparql, so each SPARQLConstraint needs # its own shape-level sh:severity. # Merged into one shape (both target Organization at Info severity) - # grouping by (target class, severity) rather than by individual check # produces identical findings with one fewer shape; see the note above # TeacherAgeQualityShape for why severity forces sh:sparql blocks onto # a shape's own sh:severity in the first place. shapes:OrganizationDissolutionQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Organization dissolution"@en ; rdfs:comment "Info: dissolved events without a manner, and dissolutions that no successor's genealogy edge points back at."@en ; sh:targetClass martialis:Organization ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "dissolution missing manner" ; sh:message "Quality check: Organization has a dissolved event with no manner - consider adding if known (merged/split/superseded/suppressed/renamed/defunct/unknown)." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:dissolvedEvent ?d . FILTER NOT EXISTS { ?d martialis:dissolutionManner ?m } } """ ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "dissolution unmatched successor edge" ; sh:message "Quality check: Organization was dissolved by merger/supersession/rename but no other organization declares the matching merged_from/replaces/renamed_from affiliation pointing back at it - consider adding the successor's edge, or the successor entity, if known." ; sh:select """ PREFIX martialis: PREFIX diss: PREFIX affrel: SELECT $this WHERE { { $this martialis:dissolvedEvent/martialis:dissolutionManner diss:merged . BIND(affrel:merged_from AS ?expectedRel) } UNION { $this martialis:dissolvedEvent/martialis:dissolutionManner diss:superseded . BIND(affrel:replaces AS ?expectedRel) } UNION { $this martialis:dissolvedEvent/martialis:dissolutionManner diss:renamed . BIND(affrel:renamed_from AS ?expectedRel) } FILTER NOT EXISTS { ?successor martialis:hasAffiliationAssertion ?a . ?a martialis:object $this ; martialis:affiliationRelation ?expectedRel . } } """ ] . # Deliberately Info, not Warning: the successor legitimately may not # be in the database yet, so both checks above are prompts, not # defects. # The mirror of the above, from the successor's side: a merged_from/ # replaces/renamed_from claim implies the predecessor ended, so its own # record should carry a dissolved event. split_from is deliberately # excluded - the parent survives a schism by definition. shapes:OrganizationPredecessorUnendedQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Predecessor unended"@en ; rdfs:comment "Info: genealogy claims whose predecessor organization has no dissolved event."@en ; sh:targetClass martialis:AffiliationAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "genealogy predecessor not dissolved" ; sh:message "Quality check: this genealogy affiliation implies the predecessor organization ended, but it has no dissolved event - consider adding one." ; sh:select """ PREFIX martialis: PREFIX affrel: SELECT $this WHERE { $this martialis:affiliationRelation ?rel ; martialis:object ?predecessor . FILTER (?rel IN (affrel:merged_from, affrel:replaces, affrel:renamed_from)) FILTER NOT EXISTS { ?predecessor martialis:dissolvedEvent ?d } } """ ] . # Alternate Name Quality Checks ################################################################# shapes:AlternateNameQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Alternate name hygiene"@en ; rdfs:comment "Warns when alternateName values contain semicolons - should be separate values."@en ; sh:targetClass martialis:Person ; # sh:severity belongs on the NodeShape, not inside sh:sparql: pyshacl # silently ignores a severity declared on the SPARQLConstraint # itself and falls back to sh:Violation (confirmed empirically; # see TeacherAgeQualityShape/InteractionPeriodQualityShape below, # which hit the same thing). sh:severity sh:Warning ; sh:sparql [ sh:name "alternate name semicolons" ; sh:message "Quality check: alternateName contains semicolons - should be split into separate values." ; sh:prefixes [ sh:declare [ sh:prefix "schema" ; sh:namespace "https://schema.org/" ] ] ; sh:select """ SELECT $this ?name WHERE { $this schema:alternateName ?name . FILTER(CONTAINS(STR(?name), ";")) } """ ] . # Confidence Assessment Quality ################################################################# # confidenceLevel is not checked here - already `required` in # martialis.schema.json's Confidence definition, so it can never fire. shapes:ConfidenceQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Confidence rationale"@en ; rdfs:comment "Info: confidence assessments without a documented rationale."@en ; sh:targetClass martialis:Confidence ; sh:property [ sh:path martialis:confidenceRationale ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: Confidence assessment has no rationale - consider documenting reasoning." ] . # Entity-Level Evidence Coverage Quality ################################################################# # `covers` (martialis:coversProperty) is only ever legal on an entity's # own top-level `sources` (SourceRef.covers via EntityEvidenceRef in # martialis.schema.json) - it is the only per-field citation mechanism # for fields with no `sources` child of their own (name, aka, # nationality, description, notes). Nested evidence (birth.sources, # teachers[].interactions[].sources, rank[].sources, etc.) is forbidden # from carrying it by the JSON Schema, since its position already scopes # it. A plain sh:targetClass martialis:SourceAssertionRelation shape # would therefore warn on every nested EvidenceRef too, which is correct # data, not missing data - so this is a SPARQLConstraint targeting the # entity itself (Person/Style/Organization/Place/Event) and checking # only its own one-hop hasEvidence, mirroring TeacherAgeQualityShape's # pattern above. Reports once per entity that has at least one # uncovered entity-level source, rather than once per uncovered source - # aggregate per-field percentages are 'martialis provenance'/'dashboard' # territory (see the file header note), this is a pointer to go look. shapes:EntityEvidenceCoverageQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Entity evidence coverage"@en ; rdfs:comment "Warns when a top-level entity source records no covers - which fields it backs is unknown."@en ; sh:targetClass martialis:Person, martialis:Style, martialis:Organization, martialis:Place, martialis:Event ; sh:severity sh:Warning ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "source missing covers" ; sh:message "Quality check: entity has a top-level source with no `covers` - consider adding which fields it backs." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:hasEvidence ?ev . ?ev a martialis:SourceAssertionRelation . FILTER NOT EXISTS { ?ev martialis:coversProperty ?p } } """ ] . # Event Completeness ################################################################# shapes:EventLocationQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Event location"@en ; rdfs:comment "Info: events with no location (place_id or geo)."@en ; sh:targetClass martialis:Event ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "event missing location" ; sh:message "Quality check: Event has no location (place_id or geo) - consider adding if known." ; sh:select """ PREFIX martialis: PREFIX schema: SELECT $this WHERE { FILTER NOT EXISTS { $this schema:location ?loc } } """ ] . shapes:ParticipationEvidenceQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Participation evidence"@en ; rdfs:comment "Warns on participation claims without their own evidence."@en ; sh:targetClass martialis:Participation ; sh:severity sh:Warning ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "participation missing sources" ; sh:message "Quality check: Participation has no sources - a participation claim (who was there, in what role) should carry its own evidence." ; sh:select """ PREFIX martialis: SELECT $this WHERE { FILTER NOT EXISTS { $this martialis:hasEvidence ?e } } """ ] . # Source Completeness ################################################################# shapes:SourceCompletenessShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Source completeness"@en ; rdfs:comment "Info: sources with no issued date."@en ; sh:targetClass martialis:Source ; sh:property [ sh:path schema:datePublished ; sh:minCount 1 ; sh:severity sh:Info ; sh:message "Quality check: Source has no issued date - consider adding if available." ] . # The remaining Source quality checks are SPARQLConstraints, each needing # its own severity. In this pipeline's pyshacl version, sh:severity placed # inside an ad-hoc sh:sparql [...] constraint blank node is not honoured - # severity must instead sit on the enclosing NodeShape (as # EntityEvidenceCoverageQualityShape above already does). Grouped here by # (target class, severity) into two shapes rather than one shape per # check - same findings, fewer shapes. shapes:SourceWebpageQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Webpage source quality"@en ; rdfs:comment "Warns on webpage sources with no accessed date or no archive snapshot."@en ; sh:targetClass martialis:Source ; sh:severity sh:Warning ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "webpage missing accessed date" ; sh:message "Quality check: webpage Source has no accessed date - when was this consulted?" ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:sourceType "webpage" . FILTER NOT EXISTS { $this martialis:accessed ?a } } """ ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "webpage missing archive snapshot" ; sh:message "Quality check: webpage Source has no archive snapshot - the live URL may go dead." ; sh:select """ PREFIX schema: PREFIX martialis: SELECT $this WHERE { $this martialis:sourceType "webpage" . FILTER NOT EXISTS { $this schema:archivedAt ?a } } """ ] . # Policy behind the containment check below: martialis:partOf is decided # by who actually published a work, not by which domain currently hosts # it - a work can be mirrored, syndicated, or archived on a third-party # site while still belonging structurally to its original outlet (e.g. # an Aikido Journal interview mirrored on a fan wiki is still part_of # Aikido Journal, not part_of the wiki, and it is not part_of anything # at all just because its own url happens to sit on a different host). # That check is the durable version of a real migration-time gap: # several items had a publisher name that already named an existing # container outlet, but no martialis:partOf link, because the original # migration matched by hostname rather than by publisher. shapes:SourceContainmentQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Source containment"@en ; rdfs:comment "Info: item-level sources without partOf, over-deep containment chains, self_published disagreement with the derivable signal, and publisher-name/container mismatches."@en ; sh:targetClass martialis:Source ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "item source missing container" ; sh:message "Quality check: an item-level source (not a container form like website/periodical/collection/book) has no part_of - consider linking it to its journal, website edition, or volume." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:sourceType ?type . FILTER (?type NOT IN ("website", "periodical", "collection", "book")) FILTER NOT EXISTS { $this martialis:partOf ?container } } """ ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "container nesting too deep" ; sh:message "Quality check: part_of nests more than 3 levels deep - worth checking this containment chain makes sense." ; sh:select """ PREFIX martialis: SELECT $this WHERE { $this martialis:partOf ?p1 . ?p1 martialis:partOf ?p2 . ?p2 martialis:partOf ?p3 . ?p3 martialis:partOf ?p4 . } """ ] ; sh:sparql [ a sh:SPARQLConstraint ; # publisher.org_id can only ever be an Organization IRI (schema: # publisher has no person-typed analogue), so this only ever # compares against a contributor's own org_id - there is no # symmetric "contributor is a Person who is also the publisher" # case to check, since publisher isn't Person-typed. An earlier # revision of this check's message overclaimed a person_id path # that was never implemented; the message here now says only # what the query actually does. sh:name "self-published signal mismatch" ; sh:message "Quality check: self_published disagrees with the derivable signal (publisher.org_id matching a contributor's own org_id) - worth double-checking." ; sh:select """ PREFIX schema: PREFIX martialis: SELECT $this WHERE { $this martialis:selfPublished ?asserted . OPTIONAL { $this schema:publisher [ martialis:publisherOrg ?pubOrg ] ; martialis:hasContributor [ martialis:contributorOrg ?pubOrg ] . BIND(true AS ?derived) } FILTER ( (?asserted = true && !BOUND(?derived)) || (?asserted = false && BOUND(?derived)) ) } """ ] ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "publisher container missed link" ; sh:message "Quality check: publisher name matches an existing, in-use container Source, but this Source has no part_of - possibly missed during migration/curation." ; sh:select """ PREFIX schema: PREFIX martialis: SELECT $this WHERE { $this schema:publisher [ schema:name ?pubName ] . FILTER NOT EXISTS { $this martialis:partOf ?anyC } ?container a martialis:Source ; schema:name ?contTitle . FILTER (?container != $this) FILTER EXISTS { ?anyItem martialis:partOf ?container } FILTER (CONTAINS(LCASE(?pubName), LCASE(?contTitle))) } """ ] . # Confidence Discrepancy ################################################################# # Migrated from bin/martialis's derive_confidence_scores/compare_confidence # (previously reported only by 'martialis provenance', with no fixture # coverage and no per-item message). The point system - type primary +2/ # secondary +1, supports true +1/false -1, weight high +1/low -1, summed # over each DISTINCT value present per source dimension (repetition adds # no points), banded (<=0 -> 0, <=2 -> 1, <=4 -> 2, else 3), then capped # at 1 if any evidence disputes (supports false) - is spelled out in full # in docs/VALIDATION.md's Analytics section, which remains authoritative; # this shape is a straight SPARQL port; see that doc for the design # rationale (why distinct-value scoring, why the disputed cap, the # GEDCOM-style "Disputed" precedent). # # Each source dimension (type/supports/weight) is scored in its own # GROUP BY subquery over a DISTINCT-value pre-aggregation, then combined # via OPTIONAL + COALESCE at the outer level - joining three independent # one-to-many subqueries directly (without pre-aggregating each to one # row per $this first) would cross-join and multiply the sums. The final # ABS(...) comparison is bound via BIND before the FILTER, not computed # inline inside the FILTER expression - verified (against a full # side-by-side run against every one of this database's assertions) # that inlining it silently drops solutions where $this has no evidence # at all, apparently a type-coercion quirk in how rdflib's SPARQL engine # evaluates a FILTER's own arithmetic subtraction against a variable # that was itself bound via a chain of nested IF()s, as opposed to a # separately pre-bound variable of the same value - pre-binding via BIND # sidesteps it entirely and is confirmed to reproduce derive_confidence_scores' # output exactly (same 124 flagged assertions, out of 1345 with both a # manual confidenceLevel and a computable derived score, on this # database's 2026-09 snapshot). shapes:ConfidenceDiscrepancyQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Confidence discrepancy"@en ; rdfs:comment "Signals assertions whose manual confidence level differs from a source-evidence-derived level by two or more."@en ; sh:targetClass martialis:TeacherAssertion, martialis:RankAssertion, martialis:HonorAssertion, martialis:NegativeAssertion, martialis:ContextualTrainingAssertion, martialis:StyleRelationAssertion, martialis:DocumentTransmissionAssertion, martialis:IdentityAssertion, martialis:AffiliationAssertion ; sh:severity sh:Info ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "confidence discrepancy" ; sh:message "Quality check: manual confidence differs from the source-evidence-derived level by 2 or more - worth double-checking." ; sh:select """ PREFIX martialis: PREFIX xsd: PREFIX srctype: PREFIX evweight: SELECT $this WHERE { $this martialis:hasConfidence [ martialis:confidenceLevel ?manual ] . OPTIONAL { SELECT $this (SUM(?tp) AS ?typeScore) WHERE { { SELECT DISTINCT $this ?type WHERE { $this (martialis:hasEvidence | martialis:hasInteraction/martialis:hasEvidence) ?ev . ?ev martialis:sourceRefType ?type . } } BIND(IF(?type = srctype:primary, 2, IF(?type = srctype:secondary, 1, 0)) AS ?tp) } GROUP BY $this } OPTIONAL { SELECT $this (SUM(?sp) AS ?supportsScore) (MAX(?df) AS ?disputedFlag) WHERE { { SELECT DISTINCT $this ?supports WHERE { $this (martialis:hasEvidence | martialis:hasInteraction/martialis:hasEvidence) ?ev . ?ev martialis:supports ?supports . } } BIND(IF(?supports = true, 1, IF(?supports = false, -1, 0)) AS ?sp) BIND(IF(?supports = false, 1, 0) AS ?df) } GROUP BY $this } OPTIONAL { SELECT $this (SUM(?wp) AS ?weightScore) WHERE { { SELECT DISTINCT $this ?weight WHERE { $this (martialis:hasEvidence | martialis:hasInteraction/martialis:hasEvidence) ?ev . ?ev martialis:evidenceWeight ?weight . } } BIND(IF(?weight = evweight:high, 1, IF(?weight = evweight:low, -1, 0)) AS ?wp) } GROUP BY $this } BIND((COALESCE(?typeScore, 0) + COALESCE(?supportsScore, 0) + COALESCE(?weightScore, 0)) AS ?score) BIND(IF(?score <= 0, 0, IF(?score <= 2, 1, IF(?score <= 4, 2, 3))) AS ?bandedLevel) BIND(IF(COALESCE(?disputedFlag, 0) = 1 && ?bandedLevel > 1, 1, ?bandedLevel) AS ?derived) BIND(ABS(xsd:integer(STR(?manual)) - xsd:integer(STR(?derived))) AS ?diff) FILTER(?diff >= 2) } """ ] . # Style Relation Integrity ################################################################# # Migrated from bin/martialis's cmd_styles, which used to detect these # by reading database/*.yaml directly (before RDF conversion). Dangling # style_id references and self-references were ALSO checked there, but # both are redundant with existing structural checks - a dangling # style_id becomes an IRI failing StyleRelationAssertionShape's # sh:class martialis:Style on martialis:object; self-references are now # StyleRelationAssertionShape's own "style relation not self-referential" # SPARQL constraint (structural, mirroring the identity/affiliation # self-reference checks already there) - so only the two genuinely # semantic checks move here. shapes:StyleSiblingAsymmetryQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Asymmetric sibling_style_of"@en ; rdfs:comment "sibling_style_of is a peer relationship and should be declared on both styles; flags a one-way declaration."@en ; sh:targetClass martialis:StyleRelationAssertion ; sh:severity sh:Warning ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "asymmetric sibling_style_of" ; sh:message "Quality check: sibling_style_of is declared one-way - the other style should declare it back." ; sh:select """ PREFIX martialis: PREFIX stylerel: SELECT $this WHERE { $this martialis:subject ?a ; martialis:object ?b ; martialis:styleRelationType stylerel:sibling_style_of . FILTER NOT EXISTS { ?rev martialis:subject ?b ; martialis:object ?a ; martialis:styleRelationType stylerel:sibling_style_of . } } """ ] . shapes:StyleVariantCycleQualityShape a sh:NodeShape ; rdfs:isDefinedBy ; rdfs:label "Style variant_of cycle"@en ; rdfs:comment "variant_of describes a narrower-to-broader lineage and should never cycle back to its own starting style."@en ; sh:targetClass martialis:Style ; sh:severity sh:Warning ; sh:sparql [ a sh:SPARQLConstraint ; sh:name "variant_of cycle" ; sh:message "Quality check: this style's variant_of chain cycles back to itself." ; sh:select """ PREFIX martialis: SELECT $this WHERE { FILTER EXISTS { $this martialis:variantOf+ $this } } """ ] .