Skip to content

Commit 200f2f0

Browse files
dgarijodgarijo
committed
Update to give local URIs to SVOs
1 parent 81f973e commit 200f2f0

File tree

1 file changed

+46
-40
lines changed

1 file changed

+46
-40
lines changed

‎CSVToRDF/src/main/java/edu/isi/csvtordf/CSV2RDF.java‎

Lines changed: 46 additions & 40 deletions
Original file line numberDiff line numberDiff line change
@@ -21,11 +21,8 @@
2121
import org.apache.jena.ontology.OntModel;
2222
import org.apache.jena.ontology.OntProperty;
2323
import org.apache.jena.ontology.OntResource;
24-
import org.apache.jena.query.Query;
25-
import org.apache.jena.query.QueryExecution;
26-
import org.apache.jena.query.QueryExecutionFactory;
27-
import org.apache.jena.query.QueryFactory;
28-
import org.apache.jena.query.ResultSet;
24+
import org.apache.jena.query.*;
25+
import org.apache.jena.rdf.model.Literal;
2926
import org.apache.jena.rdf.model.ModelFactory;
3027
import org.apache.jena.rdf.model.Property;
3128
import org.apache.jena.vocabulary.XSD;
@@ -35,6 +32,7 @@
3532
*/
3633
public class CSV2RDF {
3734
OntModel instances;
35+
OntModel variables;
3836
OntModel mcOntology;
3937
String instance_URI = "https://w3id.org/okn/i/mint/";
4038
JsonObject unitDictionary;
@@ -45,13 +43,15 @@ public class CSV2RDF {
4543
*/
4644
public CSV2RDF(String unitDictionaryFile, String scientificVariableFile) throws FileNotFoundException{
4745
instances = ModelFactory.createOntologyModel();
46+
variables = ModelFactory.createOntologyModel();
4847
//create class
4948
instances.createClass("https://w3id.org/okn/o/sd/#SoftwareConfiguration");
50-
//we read all the variables, as they will be part of the model catalog.
51-
instances.read(scientificVariableFile);
49+
//we read all the variables separately, as we will use them to link contents.
50+
variables.read(scientificVariableFile);
5251
mcOntology = ModelFactory.createOntologyModel();
5352
mcOntology.read("https://w3id.org/okn/o/sdm#");
5453
mcOntology.read("https://w3id.org/okn/o/sd#");
54+
mcOntology.createObjectProperty("http://www.w3.org/2002/07/owl#sameAs");
5555
//for some reason it's not doing the right redirect
5656
// mcOntology.read("https://mintproject.github.io/Mint-ModelCatalog-Ontology/release/1.4.0/ontology.xml");
5757
// mcOntology.read("http://ontosoft.org/ontology/software/ontosoft-v1.0.owl");//for some reason, it doesn't do the negotiation rihgt.
@@ -106,29 +106,31 @@ private void checkFile (String path){
106106
* @param rowValue
107107
*/
108108
private void processSVO(Individual ind, Property p, String rowValue, boolean index){
109+
Individual mint_svo = instances.createClass("https://w3id.org/okn/o/sd#StandardVariable").createIndividual(instance_URI+encode(rowValue));
110+
mint_svo.addLabel(rowValue, null);
111+
//assign the variable to target node.
112+
ind.addProperty(p, mint_svo);
109113
/**
110114
read label, try to find one in the svo file (there should be).
111-
* If found, then use that URI. IF not found, then create one.
115+
* If found, then copy description and label into instance, and add owl:sameAs.
112116
**/
113-
Query query = QueryFactory.create("select ?var where {"
114-
+ "?var <http://www.w3.org/2000/01/rdf-schema#label> \""+rowValue+"\"}");
115-
// Execute the query and obtain results
116-
QueryExecution qe = QueryExecutionFactory.create(query, instances);
117-
ResultSet rs = qe.execSelect();
118-
if(rs.hasNext()){
119-
ind.addProperty(p, rs.next().getResource("?var"));
120-
}else{
121-
//no variable found: create variable, add label
122-
Individual userInstance;
123-
//if(index){
124-
// userInstance = instances.createClass("https://w3id.org/okn/o/sdm#NumericalIndex").createIndividual(instance_URI+encode(rowValue));
125-
//}else{
126-
userInstance = instances.createClass("http://www.geoscienceontology.org/svo/svu#Variable").createIndividual(instance_URI+encode(rowValue));
127-
//}
128-
userInstance.addLabel(rowValue, null);
129-
//assign the variable to target node.
130-
ind.addProperty(p, userInstance);
131-
}
117+
Query query = QueryFactory.create("select ?var ?desc where {"
118+
+ "?var <http://www.w3.org/2000/01/rdf-schema#label> \""+rowValue+"\"" +
119+
"optional {?var <https://schema.org/description> ?desc}}");
120+
QueryExecution qe = QueryExecutionFactory.create(query, variables);
121+
ResultSet rs = qe.execSelect();
122+
// if found in SVO, copy description and owl:sameAs
123+
if(rs.hasNext()){
124+
QuerySolution qs = rs.next();
125+
OntProperty pSameAs = (OntProperty)mcOntology.getProperty("http://www.w3.org/2002/07/owl#sameAs");
126+
mint_svo.addProperty(pSameAs, qs.getResource("?var"));
127+
Literal desc = qs.getLiteral("?desc");
128+
if (!"".equals(desc) && desc!= null){
129+
//svo comes with html tags. remove them
130+
mint_svo.addProperty(mcOntology.getOntProperty("https://w3id.org/okn/o/sd#description"),
131+
desc.getString().replace("<p>","\n").replaceAll("\\<[^>]*>",""));
132+
}
133+
}
132134
}
133135

134136
private void processFile(String path) throws Exception{
@@ -176,10 +178,10 @@ private void processFile(String path) throws Exception{
176178
System.err.println("ERROR: "+rowValue+" is not a URI. Ind: "+ind.getLocalName());
177179
}
178180
break;
179-
case "http://www.w3.org/2001/XMLSchema#int":
181+
case "http://www.w3.org/2001/XMLSchema#integer":
180182
try{
181183
Integer.parseInt(rowValue);
182-
ind.addProperty(p, rowValue,XSDDatatype.XSDint);
184+
ind.addProperty(p, rowValue,XSDDatatype.XSDinteger);
183185
}catch (NumberFormatException e){
184186
System.err.println("ERROR: "+rowValue+" is not an Integer. Ind: "+ind.getLocalName());
185187
}
@@ -189,7 +191,7 @@ private void processFile(String path) throws Exception{
189191
Float.parseFloat(rowValue);
190192
ind.addProperty(p, rowValue,XSDDatatype.XSDfloat);
191193
}catch (NumberFormatException e){
192-
System.err.println("ERROR: "+rowValue+" is not an Integer. Ind: "+ind.getLocalName());
194+
System.err.println("ERROR: "+rowValue+" is not a float. Ind: "+ind.getLocalName());
193195
}
194196
break;
195197
default:
@@ -248,9 +250,6 @@ private void processFile(String path) throws Exception{
248250
}else if(p.toString().contains("hasStandardVariable")){
249251
processSVO(ind, p, rowValue,false);
250252
}
251-
//else if(p.toString().contains("usefulForCalculatingIndex")){
252-
// processSVO(ind, p, rowValue,true);
253-
//}
254253
else if(p.toString().contains("hasSoftwareImage")){
255254
//seaparated because here we will do the link to Dockerpedia when appropriate.
256255
//at the moment just create a URI and link with label
@@ -366,10 +365,16 @@ public static void main(String[] args){
366365
// String pathToTransformationsDataFolder = "C:\\Users\\dgarijo\\Documents\\GitHub\\ModelCatalog\\Data\\Transformations";
367366

368367
//MINT
369-
String version = "1.6.0";
370-
String pathToInstancesDataFolder = "C:\\Users\\dgarijo\\Documents\\GitHub\\ModelCatalog\\Data\\MINT\\"+version;
368+
String version = "1.6_to_1.7";
369+
// Unix
370+
String pathToGitFolder = "/home/dgarijo/Documents/GitHub/ModelCatalog";
371+
372+
// WIN
373+
// String pathToGitFolder = "C:\\Users\\dgarijo\\Documents\\GitHub";
374+
String pathToInstancesDataFolder = pathToGitFolder+"/Data/MINT/"+version;
371375
String [] graphs = {"mint@isi.edu"};//, "texas@isi.edu", "coertel@mitre.org", "brandon@starsift.com","hvarg@isi.edu"};
372-
//end mint
376+
//end MINT
377+
373378
//COVID models example
374379
// pathToInstancesDataFolder = "C:\\Users\\dgarijo\\Documents\\GitHub\\ModelCatalog\\Data\\COVID";
375380
// processDataFolder(pathToInstancesDataFolder, false, catalog);
@@ -382,11 +387,12 @@ public static void main(String[] args){
382387
//end wifire
383388
for (String graph:graphs){
384389
//a new catalog should be made per graph; otherwise graphs will be aggregated.
385-
CSV2RDF catalog = new CSV2RDF("C:\\Users\\dgarijo\\Documents\\GitHub\\ModelCatalog\\Data\\Units\\dict.json",
386-
"C:\\Users\\dgarijo\\Documents\\GitHub\\ModelCatalog\\Data\\SVO\\variable-23-10-2019.ttl");
387-
processDataFolder(pathToInstancesDataFolder+"\\"+graph, false, catalog);
388-
exportRDFFile("modelCatalog_"+graph+".ttl", catalog.instances, "TTL");
390+
CSV2RDF catalog = new CSV2RDF(pathToGitFolder+"/Data/Units/dict.json",
391+
pathToGitFolder+"/Data/SVO/variable-23-10-2019.ttl");
392+
processDataFolder(pathToInstancesDataFolder+"/"+graph, false, catalog);
393+
exportRDFFile("modelCatalog_"+graph+"_"+version+".ttl", catalog.instances, "TTL");
389394
}
395+
System.out.println("Import process finished successfully");
390396
}catch (Exception e){
391397
System.err.println("Error: "+e);
392398
}

0 commit comments

Comments
 (0)