mirror of
https://github.com/ISEAL-Community/iseal-core.git
synced 2024-11-22 06:45:02 +01:00
Update rdf_creation.py
This commit is contained in:
Marie-Angélique Laporte
parent
1b6b75f886
commit
355fa3a56f
@ -7,7 +7,6 @@ from rdflib import URIRef, BNode, Literal
|
||||
import json
|
||||
import os
|
||||
import pandas as pd
|
||||
import pylode
|
||||
import re
|
||||
|
||||
|
||||
@ -21,23 +20,23 @@ iseal = URIRef(NS)
|
||||
g.add((iseal, RDF.type, OWL.Ontology))
|
||||
|
||||
|
||||
df = pd.read_csv('../data/schema-fields.csv')
|
||||
df.dropna(how='all', axis=1)
|
||||
df.fillna('', inplace=True)
|
||||
df = pd.read_csv("../data/schema-fields.csv")
|
||||
df.dropna(how="all", axis=1)
|
||||
df.fillna("", inplace=True)
|
||||
|
||||
|
||||
for index, row in df.iterrows():
|
||||
element_name = row['element name']
|
||||
element_description = row['element description']
|
||||
comment = row['element guidance']
|
||||
example = row['element link for more information']
|
||||
cardinality = row['element options']
|
||||
prop_type = row['element type']
|
||||
controlled_vocab = row['element controlled values or terms']
|
||||
module = row['idss element cluster']
|
||||
module_cat = row['idss schema module']
|
||||
dc = row['element link for dublin core attributes']
|
||||
dspace = row['dspace field name']
|
||||
element_name = row["element name"]
|
||||
element_description = row["element description"]
|
||||
comment = row["element guidance"]
|
||||
example = row["element link for more information"]
|
||||
cardinality = row["element options"]
|
||||
prop_type = row["element type"]
|
||||
controlled_vocab = row["element controlled values or terms"]
|
||||
module = row["idss element cluster"]
|
||||
module_cat = row["idss schema module"]
|
||||
dc = row["element link for dublin core attributes"]
|
||||
dspace = row["dspace field name"]
|
||||
|
||||
##module
|
||||
moduleUri = URIRef(NS + module)
|
||||
@ -85,7 +84,6 @@ for index, row in df.iterrows() :
|
||||
|
||||
element = element_name_safe
|
||||
|
||||
|
||||
conceptUri = URIRef(NS + concept.replace(" ", "_"))
|
||||
if not (None, SKOS.prefLabel, Literal(concept)) in g:
|
||||
##create concept as skos concept
|
||||
@ -95,19 +93,37 @@ for index, row in df.iterrows() :
|
||||
|
||||
## create properties
|
||||
elementURI = URIRef(NS + element.replace(" ", "_"))
|
||||
if prop_type == 'CONTROLLED VALUE': ## object property
|
||||
if prop_type == "CONTROLLED VALUE": ## object property
|
||||
g.add((elementURI, SKOS.prefLabel, Literal(element)))
|
||||
g.add((elementURI, RDF.type, OWL.ObjectProperty))
|
||||
g.add((elementURI, OWL.domain, conceptUri))
|
||||
## add suproperty link
|
||||
if(dc):
|
||||
if dc:
|
||||
dct = dc.split(":")[1]
|
||||
if 'wgs84' in dc:
|
||||
g.add((elementURI, RDFS.subPropertyOf, URIRef("http://www.w3.org/2003/01/geo/wgs84_pos#"+dct)))
|
||||
if "wgs84" in dc:
|
||||
g.add(
|
||||
(
|
||||
elementURI,
|
||||
RDFS.subPropertyOf,
|
||||
URIRef("http://www.w3.org/2003/01/geo/wgs84_pos#" + dct),
|
||||
)
|
||||
)
|
||||
else:
|
||||
g.add((elementURI, RDFS.subPropertyOf, URIRef("http://purl.org/dc/terms/"+dct)))
|
||||
g.add(
|
||||
(
|
||||
elementURI,
|
||||
RDFS.subPropertyOf,
|
||||
URIRef("http://purl.org/dc/terms/" + dct),
|
||||
)
|
||||
)
|
||||
## add dspace alternative ID
|
||||
g.add((elementURI, URIRef("http://purl.org/dc/terms/alternative"), Literal(dspace)))
|
||||
g.add(
|
||||
(
|
||||
elementURI,
|
||||
URIRef("http://purl.org/dc/terms/alternative"),
|
||||
Literal(dspace),
|
||||
)
|
||||
)
|
||||
## create controlled vocab
|
||||
cvURI = URIRef(NS + "VOCAB_" + element.replace(" ", "_"))
|
||||
g.add((cvURI, RDF.type, OWL.Class)) ## SKOS.Concept ## SKOS.Collection??
|
||||
@ -120,10 +136,20 @@ for index, row in df.iterrows() :
|
||||
g.add((elementURI, OWL.range, cvURI))
|
||||
|
||||
## add the controlled vocab information on properties directly
|
||||
g.add((elementURI, URIRef("http://purl.org/dc/dcam/rangeIncludes"), Literal("https://raw.githubusercontent.com/alanorth/iseal-schema/main/data/controlled-vocabularies/"+element+".txt")))
|
||||
g.add(
|
||||
(
|
||||
elementURI,
|
||||
URIRef("http://purl.org/dc/dcam/rangeIncludes"),
|
||||
Literal(
|
||||
"https://raw.githubusercontent.com/alanorth/iseal-schema/main/data/controlled-vocabularies/"
|
||||
+ element
|
||||
+ ".txt"
|
||||
),
|
||||
)
|
||||
)
|
||||
|
||||
## cardinality
|
||||
if cardinality == 'MULTI SELECT FROM CONTROL LIST':
|
||||
if cardinality == "MULTI SELECT FROM CONTROL LIST":
|
||||
br = BNode()
|
||||
g.add((br, RDF.type, OWL.Restriction))
|
||||
g.add((br, OWL.onProperty, elementURI))
|
||||
@ -147,26 +173,44 @@ for index, row in df.iterrows() :
|
||||
g.add((elementURI, SKOS.prefLabel, Literal(element)))
|
||||
g.add((elementURI, RDF.type, OWL.DatatypeProperty))
|
||||
g.add((elementURI, OWL.domain, conceptUri))
|
||||
if(dc):
|
||||
if dc:
|
||||
dct = dc.split(":")[1]
|
||||
if 'wgs84' in dc:
|
||||
g.add((elementURI, RDFS.subPropertyOf, URIRef("http://www.w3.org/2003/01/geo/wgs84_pos#"+dct)))
|
||||
if "wgs84" in dc:
|
||||
g.add(
|
||||
(
|
||||
elementURI,
|
||||
RDFS.subPropertyOf,
|
||||
URIRef("http://www.w3.org/2003/01/geo/wgs84_pos#" + dct),
|
||||
)
|
||||
)
|
||||
else:
|
||||
g.add((elementURI, RDFS.subPropertyOf, URIRef("http://purl.org/dc/terms/"+dct)))
|
||||
g.add(
|
||||
(
|
||||
elementURI,
|
||||
RDFS.subPropertyOf,
|
||||
URIRef("http://purl.org/dc/terms/" + dct),
|
||||
)
|
||||
)
|
||||
## add dspace alternative ID
|
||||
g.add((elementURI, URIRef("http://purl.org/dc/terms/alternative"), Literal(dspace)))
|
||||
g.add(
|
||||
(
|
||||
elementURI,
|
||||
URIRef("http://purl.org/dc/terms/alternative"),
|
||||
Literal(dspace),
|
||||
)
|
||||
)
|
||||
range = None
|
||||
if prop_type == 'DATE':
|
||||
if prop_type == "DATE":
|
||||
g.add((elementURI, OWL.range, XSD.date))
|
||||
range = XSD.date
|
||||
elif prop_type == 'NUMERIC VALUE':
|
||||
elif prop_type == "NUMERIC VALUE":
|
||||
g.add((elementURI, OWL.range, XSD.float))
|
||||
range = XSD.float
|
||||
else:
|
||||
g.add((elementURI, OWL.range, XSD.string))
|
||||
range = XSD.string
|
||||
##cardinality
|
||||
if cardinality == 'REPEAT VALUES':
|
||||
if cardinality == "REPEAT VALUES":
|
||||
br = BNode()
|
||||
g.add((br, RDF.type, OWL.Restriction))
|
||||
g.add((br, OWL.onProperty, elementURI))
|
||||
@ -180,7 +224,6 @@ for index, row in df.iterrows() :
|
||||
g.add((br, OWL.onDataRange, range))
|
||||
g.add((conceptUri, RDFS.subClassOf, br))
|
||||
|
||||
|
||||
if comment:
|
||||
g.add((elementURI, SKOS.scopeNote, Literal(comment)))
|
||||
if example:
|
||||
@ -191,6 +234,4 @@ for index, row in df.iterrows() :
|
||||
# print(element_name)
|
||||
|
||||
## save graph
|
||||
g.serialize(destination='idds_new3.ttl', format='turtle')
|
||||
|
||||
|
||||
g.serialize(destination="idds_new3.ttl", format="turtle")
|
||||
|
||||
Loading…
Reference in New Issue
Block a user