Mercurial > hg > duomoOWLProject
annotate src/de/mpiwg/dwinter/duomo/lexdump/LexOWLTransformer.java @ 9:4392a6adf85a default tip
new version der label mit language tag
| author | dwinter |
|---|---|
| date | Thu, 16 Aug 2012 11:40:17 +0200 |
| parents | 919e9f3b5efd |
| children |
| rev | line source |
|---|---|
| 0 | 1 package de.mpiwg.dwinter.duomo.lexdump; |
| 2 | |
| 1 | 3 import java.io.File; |
| 4 import java.io.FileWriter; | |
| 0 | 5 import java.io.IOException; |
| 6 import java.net.URI; | |
| 7 import java.net.URISyntaxException; | |
| 8 import java.net.URLEncoder; | |
| 9 import java.util.Collection; | |
| 10 import java.util.HashMap; | |
| 11 import java.util.HashSet; | |
| 12 import java.util.Iterator; | |
| 13 import java.util.List; | |
| 14 import java.util.Map; | |
| 15 import java.util.Set; | |
| 16 | |
| 17 import javax.print.URIException; | |
| 18 | |
| 19 import org.apache.log4j.Level; | |
| 20 import org.apache.log4j.Logger; | |
| 21 import org.jdom.Attribute; | |
| 22 import org.jdom.Element; | |
| 23 import org.jdom.JDOMException; | |
| 24 import org.jdom.xpath.XPath; | |
| 25 | |
| 26 import com.sun.xml.internal.ws.developer.MemberSubmissionEndpointReference.Elements; | |
| 27 | |
| 28 import edu.stanford.smi.protege.exception.OntologyLoadException; | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
29 import edu.stanford.smi.protege.model.Facet; |
| 0 | 30 import edu.stanford.smi.protege.model.Instance; |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
31 import edu.stanford.smi.protege.model.Slot; |
| 4 | 32 import edu.stanford.smi.protegex.owl.jena.JenaOWLModel; |
| 0 | 33 import edu.stanford.smi.protegex.owl.model.OWLAllValuesFrom; |
| 34 import edu.stanford.smi.protegex.owl.model.OWLClass; | |
| 35 import edu.stanford.smi.protegex.owl.model.OWLDataRange; | |
| 36 import edu.stanford.smi.protegex.owl.model.OWLIndividual; | |
| 37 import edu.stanford.smi.protegex.owl.model.OWLNamedClass; | |
| 38 import edu.stanford.smi.protegex.owl.model.OWLProperty; | |
| 39 import edu.stanford.smi.protegex.owl.model.OWLUnionClass; | |
| 40 import edu.stanford.smi.protegex.owl.model.RDFList; | |
| 41 import edu.stanford.smi.protegex.owl.model.RDFProperty; | |
| 42 import edu.stanford.smi.protegex.owl.model.RDFResource; | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
43 import edu.stanford.smi.protegex.owl.model.RDFSClass; |
| 0 | 44 import edu.stanford.smi.protegex.owl.model.impl.AbstractOWLQuantifierRestriction; |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
45 import edu.stanford.smi.protegex.owl.model.impl.AbstractOWLRestriction; |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
46 import edu.stanford.smi.protegex.owl.model.impl.DefaultOWLAllValuesFrom; |
| 0 | 47 import edu.stanford.smi.protegex.owl.model.impl.DefaultOWLUnionClass; |
| 48 | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
49 /** |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
50 * @author dwinter |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
51 * |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
52 *Transformiert die Lex Files in OWL |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
53 */ |
| 0 | 54 public class LexOWLTransformer { |
| 55 private Logger logger = Logger.getRootLogger(); | |
| 1 | 56 private FileWriter missing; |
| 0 | 57 private OWLImporter owlDoc; |
| 58 private LexDumpImporter lexDoc; | |
| 59 private Map<String, OWLIndividual> individualIds = new HashMap<String, OWLIndividual>(); // speichere | |
| 60 // ids | |
| 61 // fuer | |
| 62 // weitere | |
| 63 // Verwendung | |
| 64 | |
| 1 | 65 public LexOWLTransformer(OWLImporter owlDoc, LexDumpImporter lexDoc) throws IOException { |
| 0 | 66 this.owlDoc = owlDoc; |
| 67 this.lexDoc = lexDoc; | |
| 2 | 68 missing = new FileWriter(new File("/tmp/missing.out")); |
| 0 | 69 } |
| 70 | |
| 71 /** | |
| 72 * @param args | |
| 73 * @throws Exception | |
| 74 */ | |
| 75 public static void main(String[] args) throws Exception { | |
| 6 | 76 System.out.println("Stack size has to be large, use at least -Xmx3048M -Xss10M"); |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
77 if (args.length<2){ |
|
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
78 System.out.println("Usage: \"/Users/dwinter/Documents/Projekte/Diss - data-mining/eclipseWorkspace/de.mpiwg.dwinter.duomo/owlInput\""+ |
|
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
79 "\"/Users/dwinter/Documents/Projekte/Diss - data-mining/eclipseWorkspace/duomoData/LEXXDUMP.xml\""); |
|
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
80 System.exit(-1); |
|
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
81 } |
|
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
82 |
| 0 | 83 Logger.getRootLogger().setLevel(Level.DEBUG); |
| 84 // Import OWL | |
| 85 //String base = "/Users/dwinter/Documents/Projekte/Diss - data-mining/eclipseWorkspace/de.mpiwg.dwinter.duomo/owlInput"; | |
| 86 String base=args[0]; | |
| 87 String baseUri= base.replace(" ", "%20"); | |
| 88 URI ontologieUri = new URI("file://"+baseUri+"/duomoAnalysis.owl"); | |
| 89 //URI ontologieUri = new URI( | |
| 90 // "file:///Users/dwinter/Documents/Projekte/Diss%20-%20data-mining/eclipseWorkspace/de.mpiwg.dwinter.duomo/owlInput/duomoAnalysis.owl"); | |
| 91 OWLImporter owlDoc = null; | |
| 92 try { | |
| 93 | |
| 94 owlDoc = new OWLImporter(base, ontologieUri); | |
| 95 // owlDoc.printModel(); | |
| 96 } catch (OntologyLoadException e) { | |
| 97 e.printStackTrace(); | |
| 98 System.exit(-1); | |
| 99 } | |
| 100 // read and parse lexfile | |
| 101 //String lexFile = "/Users/dwinter/Documents/Projekte/Diss - data-mining/eclipseWorkspace/duomoData/LEXXDUMP.xml"; | |
| 102 String lexFile=args[1]; | |
| 103 LexDumpImporter lexDoc = null; | |
| 104 try { | |
| 105 lexDoc = new LexDumpImporter(lexFile); | |
| 106 } catch (JDOMException e) { | |
| 107 e.printStackTrace(); | |
| 108 System.exit(-1); | |
| 109 } catch (IOException e) { | |
| 110 System.exit(-1); | |
| 111 e.printStackTrace(); | |
| 112 } | |
| 113 | |
| 114 // List<Element> cartas = lexDoc.getCartas(); | |
| 115 // System.out.println(cartas.size()); | |
| 116 | |
| 117 LexOWLTransformer tf = new LexOWLTransformer(owlDoc, lexDoc); | |
| 118 tf.transform(); | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
119 //owlDoc.save("file:///tmp/out.owl"); |
| 0 | 120 } |
| 121 | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
122 /** |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
123 * Hautmethoden zur Transformation |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
124 * @throws URISyntaxException |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
125 * @throws Exception |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
126 */ |
| 4 | 127 private void transform() throws URISyntaxException, Exception { |
| 0 | 128 List<Element> signatures = lexDoc.getSignatures(); |
| 129 | |
| 130 // Element signature = signatures.get(0); | |
| 4 | 131 int signatureCount = 0; |
| 0 | 132 int maxsign = signatures.size(); |
| 133 for (Element signature : signatures) { | |
| 134 logger.debug(String.format("Signature: %s (%s)", signatureCount, | |
| 135 maxsign)); | |
| 4 | 136 |
| 137 signatureCount++; | |
| 138 //if (signatureCount<5 || signatureCount>10) | |
| 139 // continue; | |
| 0 | 140 OWLIndividual signatureInd = createSignature(signature); |
| 141 | |
| 142 @SuppressWarnings("unchecked") | |
| 143 List<Element> cards = XPath.selectNodes(signature, ".//carta"); | |
| 144 int cardsCount = 1; | |
| 145 int maxcards = cards.size(); | |
| 146 | |
| 147 // Element card=cards.get(0); | |
| 148 for (Element card : cards) { | |
| 149 logger.debug(String.format("Cards: %s (%s)", cardsCount++, | |
| 150 maxcards)); | |
| 151 logger.debug(String.format("Signature: %s (%s)", | |
| 152 signatureCount, maxsign)); | |
| 153 OWLIndividual cardInd = createCard(card, signatureInd); | |
| 154 | |
| 155 @SuppressWarnings("unchecked") | |
| 156 List<Element> records = XPath.selectNodes(card, ".//record"); | |
| 157 for (Element record : records) { | |
| 158 // Element record = records.get(0); | |
| 159 OWLIndividual recordInd = createRecord(record, cardInd); | |
| 160 OWLIndividual recordEvent = createEvent(record, recordInd); // Event | |
| 161 // und | |
| 162 // Records | |
| 163 // sind | |
| 164 // im | |
| 165 // Original | |
| 166 // im | |
| 167 // record-tag | |
| 168 } | |
| 169 } | |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
170 //owlDoc.save("file:///tmp/out"+String.valueOf(signatureCount)+".owl"); |
|
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
171 //logger.debug("Saved:"+"/tmp/out"+String.valueOf(signatureCount)+".owl"); |
|
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
172 //owlDoc.reloadOWL(); |
| 0 | 173 } |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
174 owlDoc.printModel(); |
| 0 | 175 } |
| 176 | |
| 177 private OWLIndividual createEvent(Element record, OWLIndividual recordInd) { | |
| 178 | |
| 179 logger.debug("Create Event"); | |
| 180 OWLIndividual eventInstance = owlDoc.createInstance("RecordedEvent"); | |
| 181 | |
| 182 owlDoc.setProperty(recordInd, "crm:P70_documents", eventInstance); | |
| 183 | |
| 184 // timespan | |
| 185 | |
| 186 try { | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
187 String dateDcStart = lexDoc.getValue(record, ".//datrf/startdate"); |
| 0 | 188 |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
189 String dateDcEnd = lexDoc.getValue(record, ".//datrf/enddate"); |
| 0 | 190 |
| 191 if (!dateDcStart.equals("")) { | |
| 192 OWLIndividual timeSpan = owlDoc.createTimeSpan(dateDcStart, | |
| 193 dateDcEnd); | |
| 194 owlDoc.setProperty(eventInstance, "crm:P4_has_time-span", | |
| 195 timeSpan); | |
| 196 } | |
| 197 } catch (JDOMException e) { | |
| 198 // TODO Auto-generated catch block | |
| 199 e.printStackTrace(); | |
| 200 } | |
| 201 | |
| 202 // tipol | |
| 203 String typology; | |
| 204 try { | |
| 205 typology = lexDoc.getValue(record, ".//tipol"); | |
| 206 OWLIndividual typolInd = createOrGetTipol(typology); | |
| 207 owlDoc.setProperty(eventInstance, "has_tipol", typolInd); | |
| 208 | |
| 209 } catch (JDOMException e) { | |
| 210 // TODO Auto-generated catch block | |
| 211 e.printStackTrace(); | |
| 212 } | |
| 213 | |
| 214 // reges | |
| 215 try { | |
| 216 createNewDependingDataTypePropertyFromXpath(record, eventInstance, | |
| 217 ".//reges/italian", "has_reges", "it"); | |
| 218 createNewDependingDataTypePropertyFromXpath(record, eventInstance, | |
| 219 ".//reges/english", "has_reges", "en"); | |
| 220 } catch (JDOMException e) { | |
| 221 // TODO Auto-generated catch block | |
| 222 e.printStackTrace(); | |
| 223 } | |
| 224 | |
| 225 // nomiq | |
| 226 | |
| 227 try { | |
| 228 @SuppressWarnings("unchecked") | |
| 229 List<Element> nomiqs = XPath.selectNodes(record, ".//nomiq"); | |
| 230 OWLIndividual recordNamesRoles = null; | |
| 231 for (Element nomiq : nomiqs) { | |
| 232 String name = lexDoc.getValue(nomiq, "./name"); | |
| 233 String role = lexDoc.getValue(nomiq, "./role"); | |
| 234 | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
235 String provenance = lexDoc.getValue(nomiq, "./name/provenance"); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
236 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
237 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
238 |
| 0 | 239 if (!name.equals("") && !role.equals("")) { |
| 240 recordNamesRoles = handleNameWithRole(recordInd, name, role); | |
| 241 } else if (!role.equals("")) { | |
| 2 | 242 recordNamesRoles = createOrGetRole(role); |
| 0 | 243 } else if (!name.equals("")) { |
| 2 | 244 recordNamesRoles = createOrGetName(name); |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
245 if (provenance!=""){ |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
246 owlDoc.setDataTypePropery(recordNamesRoles, "has_provenance_as_string", provenance, "it"); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
247 } |
| 0 | 248 } |
| 249 | |
| 250 if (recordNamesRoles != null) { | |
| 251 owlDoc.setProperty(eventInstance, | |
| 252 "recordsDuomoObjectNameRoles", recordNamesRoles); | |
| 253 String id = lexDoc.getValue(nomiq, "./@id"); | |
| 254 individualIds.put(id, recordNamesRoles); | |
| 255 } | |
| 256 } | |
| 257 | |
| 258 } catch (JDOMException e) { | |
| 259 // TODO Auto-generated catch block | |
| 260 e.printStackTrace(); | |
| 261 } | |
| 262 | |
| 263 // istit | |
| 264 try { | |
| 7 | 265 String istit = lexDoc.getValue(record, ".//istit"); |
| 266 if (!istit.equals("")){ | |
| 0 | 267 OWLIndividual istitInd = owlDoc.getIndividualByReadableId( |
| 268 "IndicesInstitutions", istit); | |
| 269 if (istitInd == null) { | |
| 270 istitInd = createOrGetInstitution(istit); | |
| 271 } | |
| 7 | 272 |
| 0 | 273 |
| 274 owlDoc.setProperty(eventInstance, "recordsDuomoObjectInstitution", | |
| 275 istitInd); | |
| 7 | 276 } |
| 0 | 277 |
| 278 } catch (JDOMException e) { | |
| 279 // TODO Auto-generated catch block | |
| 280 e.printStackTrace(); | |
| 281 } | |
| 282 | |
| 283 // types | |
| 284 | |
| 285 try { | |
| 286 List<Element> types = XPath.selectNodes(record, ".//type"); | |
| 287 for (Element type : types) { | |
| 288 createType(eventInstance, type); | |
| 289 } | |
| 290 } catch (JDOMException e) { | |
| 291 // TODO Auto-generated catch block | |
| 292 e.printStackTrace(); | |
| 293 } | |
| 294 | |
| 295 return eventInstance; | |
| 296 } | |
| 297 | |
| 298 private void createType(OWLIndividual eventInstance, Element type) { | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
299 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
300 |
| 0 | 301 String typeId; |
| 302 try { | |
| 303 typeId = lexDoc.getValue(type, "./ptr/@target"); | |
| 304 String clsName = owlDoc.getClassNameFromTypeId(typeId); | |
| 305 OWLIndividual typeInd = owlDoc.createInstance(clsName); | |
| 306 | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
307 owlDoc.setProperty(eventInstance, "has_topic", typeInd); |
| 0 | 308 List<Element> freeTexts = XPath.selectNodes(type, "./freetext"); |
| 309 for (Element freeText : freeTexts) { | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
310 OWLNamedClass cls = owlDoc.getClassFromTypeId(typeId); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
311 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
312 OWLNamedClass subjectClass = getPreferredTargetClass(cls, |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
313 "has_subject"); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
314 OWLNamedClass predicateClass = getPreferredTargetClass(cls, |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
315 "has_predicate"); |
| 0 | 316 |
| 317 String subjPointer = lexDoc.getValue(freeText, | |
| 318 "./sub/ptrtoperson/@target"); | |
| 319 String subjText = lexDoc.getValue(freeText, "./sub"); | |
| 320 | |
| 321 OWLIndividual subjInd = createSubjectOrPredicate(subjectClass, | |
| 322 subjPointer, subjText); | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
323 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
324 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
325 //suche ob eine subpropery von materiaInvolved fuer die die zem Type (type) gehoerige Klasse (clsName) existiert |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
326 // und wenn ja welche, TODO: zur Zeit wird dann aus dem String "subjText" das entsprechende Material erzeugt. |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
327 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
328 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
329 //Collection<RDFProperty> props = cls.getPossibleRDFProperties(); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
330 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
331 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
332 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
333 RDFProperty superproperty= owlDoc.owlModel.getRDFProperty("http://ontologies.mpiwg-berlin.mpg.de/research/duomoAnalysis.owl/materialInvolved"); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
334 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
335 Collection<RDFSClass> sc = cls.getSuperclasses(true); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
336 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
337 OWLNamedClass mat = owlDoc.owlModel.getOWLNamedClass("http://ontologies.mpiwg-berlin.mpg.de/research/duomoAnalysis.owl/DuomoActivities"); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
338 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
339 if (sc.contains(mat)){ // cls is subclass of DuomoActivities |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
340 Collection<RDFProperty> props = superproperty.getSubproperties(false); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
341 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
342 //jetzt suche zu welcher subproperty die classe cls gehšrt |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
343 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
344 //dazu gehe durch alle subproperties von materialInvolved- |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
345 for (RDFProperty prop:props){ |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
346 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
347 @SuppressWarnings("unchecked") |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
348 Collection<RDFSClass> domains = prop.getDomains(true); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
349 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
350 for(RDFSClass domain: domains){ |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
351 //if (domain.getName().equals(cls.getName())) |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
352 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
353 //suche jetzt die domaene zu diesen property in schaue ob die cls eine subklasse davon ist |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
354 if (cls.isSubclassOf(domain)) // cls ist in der domaene der property |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
355 { |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
356 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
357 //die propery muss genau einen wert aus einer festen klasse haben, diese wird jetzt gesucht und dann eine individual erzeugt. |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
358 Collection<AbstractOWLRestriction> restrictions = cls.getRestrictions(prop, false); // suche die restriction und erzeuge dann ein object dieses type |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
359 for (AbstractOWLRestriction restriction: restrictions){ |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
360 if (DefaultOWLAllValuesFrom.class.isInstance(restriction)){ |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
361 DefaultOWLAllValuesFrom rest = (DefaultOWLAllValuesFrom)restriction; |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
362 RDFResource restClass = rest.getAllValuesFrom(); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
363 OWLIndividual inst = owlDoc.createOrGetInstanceWithIdentifier(restClass.getLocalName(), "Identifier", subjText, false); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
364 owlDoc.setProperty(typeInd, prop.getLocalName(), inst); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
365 //materialInd = owlDoc.createInstance(res.getName()); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
366 } |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
367 } |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
368 } |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
369 } |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
370 } |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
371 } |
| 0 | 372 |
| 373 String predPointer = lexDoc.getValue(freeText, | |
| 374 "./pred/ptrtoperson/@target"); | |
| 375 String predText = lexDoc.getValue(freeText, "./pred"); | |
| 376 | |
| 377 OWLIndividual predInd = createSubjectOrPredicate( | |
| 378 predicateClass, predPointer, predText); | |
| 379 | |
| 380 if (subjInd != null) { | |
| 381 owlDoc.setProperty(typeInd, "has_subject", subjInd); | |
| 382 } | |
| 383 | |
| 384 if (predInd != null) { | |
| 385 owlDoc.setProperty(typeInd, "has_predicate", predInd); | |
| 386 } | |
| 387 | |
| 388 owlDoc.setProperty(eventInstance, "recordsDuomoObjectConcept", | |
| 389 typeInd); | |
| 390 } | |
| 391 | |
| 392 } catch (JDOMException e) { | |
| 393 // TODO Auto-generated catch block | |
| 394 e.printStackTrace(); | |
| 395 } | |
| 396 | |
| 397 } | |
| 398 | |
| 399 private OWLIndividual createSubjectOrPredicate(OWLNamedClass toClass, | |
| 400 String subjPointer, String subjText) { | |
| 401 OWLIndividual subjInd = null; | |
| 402 | |
| 403 if (!subjPointer.equals("")) { | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
404 subjInd = owlDoc.createInstance(toClass.getName()); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
405 //subjInd = toClass.createOWLIndividual(null); |
| 0 | 406 OWLIndividual ind = individualIds.get(subjPointer); |
| 407 if (ind == null) { | |
| 408 logger.debug("target ID does not exist:" + subjPointer); | |
| 1 | 409 try { |
| 410 missing.write("target ID does not exist:" + subjPointer+"\n"); | |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
411 missing.flush(); |
| 1 | 412 } catch (IOException e) { |
| 413 // TODO Auto-generated catch block | |
| 414 e.printStackTrace(); | |
| 415 } | |
| 0 | 416 } else { |
| 417 owlDoc.setProperty(subjInd, "has_NameOrRoleFromIndex", ind); | |
| 418 } | |
| 419 } | |
| 420 | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
421 if (!subjText.equals("") & !subjText.equals(" ")) { |
| 0 | 422 if (subjInd == null) |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
423 subjInd = owlDoc.createInstance(toClass.getName()); |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
424 //subjInd = toClass.createOWLIndividual(null); |
| 0 | 425 |
| 426 OWLNamedClass idcls = owlDoc.owlModel | |
| 427 .getOWLNamedClass("Identifier"); // is die klasse selbst | |
| 428 // schon ein identifiert | |
| 429 if (toClass.getNamedSuperclasses(true).contains(idcls)) { // to | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
430 owlDoc.setProperty(subjInd, "rdfs:label", subjText); |
| 0 | 431 } else { |
| 432 | |
| 433 OWLIndividual ident = owlDoc | |
| 434 .createInstance("IdentifierPredicateOrSubject"); | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
435 owlDoc.setProperty(ident, "rdfs:label", subjText); |
| 0 | 436 owlDoc.setProperty(subjInd, "crm:P48_has_preferred_identifier", |
| 437 ident); | |
| 438 } | |
| 439 } | |
| 440 return subjInd; | |
| 441 } | |
| 442 | |
| 443 private OWLNamedClass getPreferredTargetClass(OWLNamedClass cls, | |
| 444 String propertyName) { | |
| 445 RDFProperty prop = owlDoc.owlModel.getRDFProperty(propertyName); | |
| 446 | |
| 447 // finde welche klasse als subject erlaubt ist | |
| 448 Collection<?> restrictions = cls.getRestrictions(prop, true); | |
| 449 | |
| 450 RDFResource restrictionValues = prop.getRange(); // nimm erstmal den | |
| 451 // gesammten Range | |
| 452 | |
| 453 // schaue jetzt nach ob eb es einschraenkungen gibt. | |
| 454 for (Iterator<?> it = restrictions.iterator(); it.hasNext();) { | |
| 455 Object restriction = it.next(); | |
| 456 if (OWLAllValuesFrom.class.isInstance(restriction)) { | |
| 457 OWLAllValuesFrom ar = (OWLAllValuesFrom) restriction; | |
| 458 restrictionValues = ar.getAllValuesFrom(); | |
| 459 break; | |
| 460 } | |
| 461 | |
| 462 } | |
| 463 OWLNamedClass toClass = null; | |
| 464 if (OWLNamedClass.class.isInstance(restrictionValues)) { | |
| 465 toClass = (OWLNamedClass) restrictionValues; | |
| 466 } else if (OWLDataRange.class.isInstance(restrictionValues)) { | |
| 467 RDFList dr = ((OWLDataRange) restrictionValues).getOneOf(); | |
| 468 for (Object d : dr.getValues()) { | |
| 469 System.out.println(d); | |
| 470 toClass = (OWLNamedClass) d; | |
| 471 // FIXME: geht das?? | |
| 472 } | |
| 473 } else if (DefaultOWLUnionClass.class.isInstance(restrictionValues)) { // mehr | |
| 474 // als | |
| 475 // eine | |
| 476 // moeglich | |
| 477 DefaultOWLUnionClass ou = (DefaultOWLUnionClass) restrictionValues; | |
| 478 Set set = new HashSet(); | |
| 479 ou.getNestedNamedClasses(set); | |
| 480 | |
| 481 for (Iterator<?> it = set.iterator(); it.hasNext();) { | |
| 482 OWLNamedClass cl = (OWLNamedClass) it.next(); | |
| 483 OWLNamedClass idcls = owlDoc.owlModel | |
| 484 .getOWLNamedClass("Identifier"); // nimm die Klasse die | |
| 485 // ein Identifier | |
| 486 // ist. | |
| 487 if (cl.getNamedSuperclasses(true).contains(idcls)) { | |
| 488 toClass = cl; | |
| 489 break; | |
| 490 } | |
| 491 | |
| 492 } | |
| 493 } | |
| 494 return toClass; | |
| 495 } | |
| 496 | |
| 2 | 497 |
| 0 | 498 |
| 499 private OWLIndividual handleNameWithRole(OWLIndividual recordInd, | |
| 500 String name, String role) { | |
| 501 // teste ob schon ein solcer Eintrag existiert | |
| 502 OWLIndividual nameInd = owlDoc.getIndividualByReadableId( | |
| 503 "IndicesNames", name); | |
| 504 Boolean createNewNameWithRole = false; | |
| 505 if (nameInd == null) { | |
| 506 nameInd = createOrGetName(name); | |
| 507 createNewNameWithRole = true; // name existierte nicht dann kann | |
| 508 // auch NameWithRole nicht | |
| 509 // existierten. | |
| 510 } | |
| 511 OWLIndividual roleInd = owlDoc.getIndividualByReadableId( | |
| 512 "IndicesRoles", role); | |
| 513 if (roleInd == null) { | |
| 2 | 514 roleInd = createOrGetRole(role); |
| 0 | 515 createNewNameWithRole = true; // role existierte nicht dann kann |
| 516 // auch NameWithRole nicht | |
| 517 // existierten. | |
| 518 } | |
| 519 OWLIndividual nameWithRoleInd = null; | |
| 520 if (!createNewNameWithRole) { // schon klar, dass er nicht existiert | |
| 521 nameWithRoleInd = getNameWithRole(nameInd, roleInd); | |
| 522 } | |
| 523 | |
| 524 if (nameWithRoleInd == null) { // existiert nicht | |
| 525 nameWithRoleInd = createNameWithRole(nameInd, roleInd); | |
| 526 } | |
| 527 | |
| 528 return nameWithRoleInd; | |
| 529 } | |
| 530 | |
| 531 private OWLIndividual getNameWithRole(OWLIndividual nameInd, | |
| 532 OWLIndividual roleInd) { | |
| 533 List<OWLIndividual> indicesWithNames = owlDoc.getIndividuals( | |
| 534 "IndicesNameWithRole", "refers_to_name", nameInd); // suche alle | |
| 535 // infrage | |
| 536 // kommenden | |
| 537 // nameen | |
| 538 | |
| 539 if (indicesWithNames == null) { // kein Treffer | |
| 540 return null; | |
| 541 } | |
| 542 | |
| 543 for (OWLIndividual name : indicesWithNames) { | |
| 544 Object role = owlDoc.getRelatedIndividual(name, "refers_to_role"); | |
| 545 if (roleInd.equals(role)) { | |
| 546 return name; | |
| 547 } | |
| 548 } | |
| 549 return null; | |
| 550 } | |
| 551 | |
| 552 private OWLIndividual createOrGetTipol(String typology) { | |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
553 OWLIndividual tipol = owlDoc.createOrGetInstanceWithIdentifier("Typology","Identifier",typology,false); |
| 0 | 554 return tipol; |
| 555 } | |
| 556 | |
| 557 private OWLIndividual createNameWithRole(OWLIndividual nameInd, | |
| 558 OWLIndividual roleInd) { | |
| 559 | |
| 560 OWLIndividual nameWithRoleInd = owlDoc | |
| 561 .createInstance("IndicesNameWithRole"); | |
| 562 | |
| 563 owlDoc.setProperty(nameWithRoleInd, "refers_to_name", nameInd); | |
| 564 owlDoc.setProperty(nameWithRoleInd, "refers_to_role", roleInd); | |
| 565 | |
| 566 return nameWithRoleInd; | |
| 567 | |
| 568 } | |
| 569 | |
| 570 private OWLIndividual createOrGetInstitution(String name) { | |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
571 OWLIndividual nameInd = owlDoc.createOrGetInstanceWithIdentifier("IndicesInstitutions","IdentifierInstitutions",name,false); |
| 0 | 572 |
| 573 return nameInd; | |
| 574 } | |
| 575 | |
| 576 private OWLIndividual createOrGetName(String name) { | |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
577 OWLIndividual nameInd = owlDoc.createOrGetInstanceWithIdentifier("IndicesNames","IdentifierNames",name,false); |
| 0 | 578 return nameInd; |
| 579 } | |
| 580 | |
| 2 | 581 private OWLIndividual createOrGetRole(String name) { |
|
5
923e73fa07f1
Minor changes, Suche nach Objekten nach readableId jetzt einschraenkbar, nicht auf subclassen ausweiten.
dwinter
parents:
4
diff
changeset
|
582 OWLIndividual roleInd = owlDoc.createOrGetInstanceWithIdentifier("IndicesRoles","IdentifierRoles",name,false); |
| 0 | 583 return roleInd; |
| 584 } | |
| 585 | |
| 586 private OWLIndividual createRecord(Element record, OWLIndividual cardInd) | |
| 587 throws JDOMException { | |
| 588 OWLIndividual recordInstance = owlDoc.createInstance("Record"); | |
| 589 owlDoc.setProperty(recordInstance, "is_on_card", cardInd); | |
| 590 createNewDependingInstanceFromXpath(record, recordInstance, "./@id", | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
591 new String[] { "rdfs:label" }, |
| 0 | 592 "IdentifierCurrent", "crm:P48_has_preferred_identifier"); |
| 593 | |
| 594 String value = lexDoc.getValue(record, ".//textblockid"); | |
| 595 if (!value.equals("")) | |
| 596 owlDoc.setProperty(recordInstance, "has_textblockid", value); | |
| 597 | |
| 598 String endOnCarta = lexDoc.getValue(record, "./@end_on_carta"); | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
599 |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
600 //FIXME: addRecordToCarta ist buggy. siehe dort! ausserdem wir nicht berŸcksichtig, dass zwischen |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
601 // card und end_on_carta mehr als eine liegen kann, zur Zeit wird nur die carta die in end_on_carta beschrieben wird zu |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
602 // record mittels is_on_card hinzugefŸgt. |
| 0 | 603 if (!endOnCarta.equals("")) { |
| 604 OWLIndividual signature = (OWLIndividual) owlDoc | |
| 605 .getRelatedIndividual(cardInd, "has_signature"); | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
606 addRecordToCarta(recordInstance, endOnCarta, signature); |
| 0 | 607 } |
| 608 | |
| 609 String dateDcStart = lexDoc.getValue(record, ".//datdc/startdate"); | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
610 String dateDcEnd = lexDoc.getValue(record, ".//datdc/enddate"); |
| 0 | 611 |
| 612 OWLIndividual timeSpan = owlDoc.createTimeSpan(dateDcStart, dateDcEnd); | |
| 613 | |
| 614 owlDoc.setProperty(recordInstance, "crm:P4_has_time-span", timeSpan); | |
| 615 | |
| 616 return recordInstance; | |
| 617 } | |
| 618 | |
| 619 private void addRecordToCarta(OWLIndividual recordInstance, String cardID, | |
| 620 OWLIndividual signature) { | |
| 621 | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
622 //FIXME: cartID ist nur innerhalb einer Signatur eindeutig, d.h. h, es muss die cardID gefunden werden die in der |
|
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
623 // selben signatur lebt wir "signature" |
| 0 | 624 OWLIndividual card = owlDoc.getIndividualByReadableId("Card", cardID); |
| 625 | |
| 626 if (card == null) { | |
| 627 card = createCard(cardID, signature); | |
| 628 } | |
| 629 | |
| 630 owlDoc.setProperty(recordInstance, "is_on_card", card); | |
| 631 } | |
| 632 | |
| 633 private OWLIndividual createSignature(Element signature) | |
| 634 throws JDOMException { | |
| 635 Element segHeaderElement = (Element) XPath.selectSingleNode(signature, | |
| 636 ".//segheader"); | |
| 637 String segheader = segHeaderElement.getTextTrim(); | |
| 638 | |
| 639 OWLIndividual signatureInstance = owlDoc.createInstance("Signatur"); | |
| 9 | 640 |
| 641 | |
| 642 OWLIndividual preferredId = owlDoc.createInstance("Identifier"); | |
| 643 owlDoc.setProperty(preferredId, "rdfs:label", segheader); | |
| 644 | |
| 645 owlDoc.setProperty(signatureInstance, "crm:P48_has_preferred_identifier", | |
| 646 preferredId); | |
| 647 | |
| 648 | |
| 0 | 649 return signatureInstance; |
| 650 } | |
| 651 | |
| 652 private OWLIndividual createCard(String cardId, OWLIndividual signature) { | |
| 653 | |
| 654 OWLIndividual cardInstance = owlDoc.createInstance("Card"); | |
| 655 | |
| 656 owlDoc.setProperty(cardInstance, "has_signature", signature); | |
| 657 | |
| 658 OWLIndividual preferredId = owlDoc.createInstance("IdentifierCurrent"); | |
| 659 owlDoc.setProperty(preferredId, "rdfs:label", cardId); | |
| 660 | |
| 661 owlDoc.setProperty(cardInstance, "crm:P48_has_preferred_identifier", | |
| 662 preferredId); | |
| 663 | |
| 664 return cardInstance; | |
| 665 } | |
| 666 | |
| 667 private OWLIndividual createCard(Element card, OWLIndividual signatureInd) { | |
| 668 | |
| 669 OWLIndividual cardInstance = owlDoc.createInstance("Card"); | |
| 670 | |
| 671 try { | |
| 672 | |
| 673 createNewDependingInstanceFromXpath(card, cardInstance, | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
674 ".//cartanr", new String[] { "rdfs:label" }, "IdentifierCurrent", |
| 0 | 675 "crm:P48_has_preferred_identifier"); |
| 676 | |
| 677 createNewDependingInstanceFromXpath(card, cardInstance, | |
|
8
919e9f3b5efd
neue klassen zur textanalyse (stanford parser eingebaut)
dwinter
parents:
7
diff
changeset
|
678 ".//cartaant", new String[] { "rdfs:label" }, "IdentifierCurrent", |
| 0 | 679 "crm:P1_is_identified_by"); |
| 680 | |
| 681 owlDoc.setProperty(cardInstance, "has_signature", signatureInd); | |
| 682 | |
| 683 } catch (JDOMException e) { | |
| 684 e.printStackTrace(); | |
| 685 return null; | |
| 686 } | |
| 687 | |
| 688 return cardInstance; | |
| 689 } | |
| 690 | |
| 691 private void createNewDependingInstanceFromXpath(Element card, | |
| 692 OWLIndividual cardInstance, String xpath, String[] propertyNames, | |
| 693 String newInstanceClassName, String relationNameToNewInstance) | |
| 694 throws JDOMException { | |
| 695 | |
| 696 List<?> identifierIdEls = (List<?>) XPath.selectNodes(card, xpath); | |
| 697 for (Object identifierIdEl : identifierIdEls) { | |
| 698 String identifierId = ""; | |
| 699 if (Element.class.isInstance(identifierIdEl)) { | |
| 700 identifierId = ((Element) identifierIdEl).getTextTrim(); | |
| 701 } else if (Attribute.class.isInstance(identifierIdEl)) { | |
| 702 identifierId = ((Attribute) identifierIdEl).getValue(); | |
| 703 } | |
| 704 | |
| 705 OWLIndividual identifier = owlDoc | |
| 706 .createInstance(newInstanceClassName); | |
| 707 for (int i = 0; i < propertyNames.length; i++) { | |
| 708 owlDoc.setProperty(identifier, propertyNames[i], identifierId); | |
| 709 } | |
| 710 | |
| 711 owlDoc.setProperty(cardInstance, relationNameToNewInstance, | |
| 712 identifier); | |
| 713 } | |
| 714 | |
| 715 } | |
| 716 | |
| 717 // createNewDependingDataTypePropertyFromXpath(record, eventInstance, | |
| 718 // ".//resges/italian", "has_reges", | |
| 719 // "italian"); | |
| 720 private void createNewDependingDataTypePropertyFromXpath(Element record, | |
| 721 OWLIndividual eventInstance, String xpath, String propertyName, | |
| 722 String lang) throws JDOMException { | |
| 723 | |
| 724 List<?> identifierIdEls = (List<?>) XPath.selectNodes(record, xpath); | |
| 725 for (Object identifierIdEl : identifierIdEls) { | |
| 726 String identifierId = ""; | |
| 727 if (Element.class.isInstance(identifierIdEl)) { | |
| 728 identifierId = ((Element) identifierIdEl).getTextTrim(); | |
| 729 } else if (Attribute.class.isInstance(identifierIdEl)) { | |
| 730 identifierId = ((Attribute) identifierIdEl).getValue(); | |
| 731 } | |
| 732 | |
| 733 owlDoc.setDataTypePropery(eventInstance, propertyName, | |
| 734 identifierId, lang); | |
| 735 } | |
| 736 } | |
| 737 } |