public Set<String> findCandidatesForClass(String classUri) {
Set<String> listOfPotentialCandidates = new HashSet<String>();
listOfPotentialCandidates.addAll(
luceneAnnotator.getDefinedPropertiesWhereClassIsADomain(classUri, forceSuperClasses));
listOfPotentialCandidates.addAll(
luceneAnnotator.getDefinedPropertiesWhereClassIsARange(classUri, forceSuperClasses));
Set<String> filteredList = new HashSet<String>();
for (String t : listOfPotentialCandidates) {
if (!isInIgnoreNameSpaceList(t)) {
filteredList.add(t);
}
}
return filteredList;
}
/**
* @param dTypes
* @return
*/
List<String> filterDirectTypes(List<String> dTypes) {
List<String> filteredList = new ArrayList();
List<String> toRemove = new ArrayList();
if (dTypes != null && dTypes.size() > 1) {
// Map<String, Set<String>> subClasses = ontology2Map.getSubClasses();
List<String> copyOfDtypes = new ArrayList();
copyOfDtypes.addAll(dTypes);
for (String classUri : dTypes) {
Set<String> subClassesOfThis =
luceneAnnotator.findSubClasses(classUri); // subClasses.get(classUri);
for (String uri : copyOfDtypes) {
if (subClassesOfThis.contains(uri) && !classUri.equals(uri)) {
// remove this
toRemove.add(uri);
}
}
// classUri subclass DTypes[i]; ostaje
// classUri superclass DTypes[i]; remove classUri
// OVO NE TREBA DA BUDE OVDE; PROBAJ DA TO UBACIS TAMO KAD TRAZIS
// SUGGESTIONS ILI RAZMISLI GDE JE VEC NAJBOLJE
}
}
filteredList.addAll(dTypes);
filteredList.removeAll(toRemove);
return filteredList;
}
/**
* finds elements which are the top most in the ontology...
*
* @return
*/
private List<OntologyElement> findGenericOntologyElements(Integer max) throws Exception {
List<OntologyElement> elements = new ArrayList<OntologyElement>();
Set<String> suggestions = luceneAnnotator.findPropertyURIs(max);
// now remove all that are after max
List<String> list = new ArrayList<String>(suggestions);
if (suggestions != null && suggestions.size() > max) {
for (int i = max; i < suggestions.size(); i++) {
String prop = list.get(i);
suggestions.remove(prop);
}
}
elements.addAll(returnPropertyElements(suggestions));
long start = System.currentTimeMillis();
Set<String> classUris = luceneAnnotator.findTopClasses();
long end = System.currentTimeMillis();
logger.info("Finished searching lucene for top classes for:" + (end - start) + "ms.");
if (classUris == null) classUris = new HashSet();
logger.info("Found " + classUris.size() + " top classes.");
// now remove all that are after max
List<String> newList = new ArrayList<String>(classUris);
if (newList.size() > max) {
for (int i = max; i < newList.size(); i++) {
String classUri = newList.get(i);
classUris.remove(classUri);
}
}
for (String euri : classUris) {
SerializableURI elementUri = null;
try {
elementUri = new SerializableURI(euri, false);
} catch (Exception e) {
e.printStackTrace();
}
OntologyElement e = new ClassElement();
e.setData(elementUri);
elements.add(e);
}
return elements;
}
/**
* find properties with no domain or range defined and also add some generic elements like label
* or type
*
* @return
*/
List<OntologyElement> findHangingElements() throws Exception {
List<OntologyElement> elements = new ArrayList<OntologyElement>();
Set<String> properties = new HashSet<String>();
properties.add(RDFS.LABEL.toString());
properties.add(RDF.TYPE.toString());
String table = suggestionsHelper.getDatatypePropertiesNoDomain();
properties.addAll(StringUtil.fromStringToSet(table));
String thingUri = "http://www.w3.org/2002/07/owl#Thing";
properties.addAll(
luceneAnnotator.getDefinedPropertiesWhereClassIsADomain(thingUri, forceSuperClasses));
properties.addAll(
luceneAnnotator.getDefinedPropertiesWhereClassIsARange(thingUri, forceSuperClasses));
elements.addAll(returnPropertyElements(properties));
logger.info("Adding:" + properties.size() + " hanging properties");
Set<String> classes = new HashSet<String>();
classes.add(thingUri);
elements.addAll(returnClassElements(classes));
return elements;
}
private List<OntologyElement> findCandidatesForADTPV(
DatatypePropertyValueElement el, String text) {
// Set<String> allSuggestions = new HashSet<String>();
long start = System.currentTimeMillis();
DatatypePropertyValueIdentifier dtpv = (DatatypePropertyValueIdentifier) el.getData();
List<SerializableURI> instanceUris = dtpv.getInstanceURIs();
Set<String> allClassUris = new HashSet<String>();
for (SerializableURI uri : instanceUris) {
List<String> dTypes = luceneAnnotator.findDirectTypes(uri.toString());
allClassUris.addAll(dTypes);
}
long end = System.currentTimeMillis();
logger.info(
"Lucene finished findDirectTypes for "
+ (end - start)
+ "ms and found:"
+ allClassUris.size()
+ " dTypes");
List<OntologyElement> elements = new ArrayList();
Set<String> properties = new HashSet();
start = System.currentTimeMillis();
for (String classUri : allClassUris) {
properties.addAll(findCandidatesForClass(classUri));
}
end = System.currentTimeMillis();
logger.info(
"findCandidatesForClass for "
+ (end - start)
+ "ms and found:"
+ properties.size()
+ " prop uris");
start = System.currentTimeMillis();
try {
elements.addAll(returnPropertyElements(properties));
} catch (Exception e) {
// TODO Auto-generated catch block
e.printStackTrace();
}
end = System.currentTimeMillis();
logger.info(
"returnPropertyElements for "
+ (end - start)
+ "ms and found:"
+ properties.size()
+ " prop uris");
return elements;
}
/**
* if element is datatype property then add min, max and sum as options
*
* @param element
* @param poc
* @return
*/
List<Vote> addAdditionalVotes(
OntologyElement element, POC poc, String text, String propertyUri, boolean alreadyAdded) {
// if element is datatype property add suggestions with min,
// max, sum
List<Vote> votes = new ArrayList<Vote>();
// logger.info("Adding additional votes(min,max,etc), poc is:"+poc.toString());
if (element instanceof PropertyElement) {
PropertyElement clonedElement = new PropertyElement();
String thisPropertyUri = ((PropertyElement) element).getData().toString();
// logger.info("thisPropertyUri "+thisPropertyUri);
if (luceneAnnotator.isItDatatypeProperty(thisPropertyUri)) {
((PropertyElement) clonedElement).setDatatypeProperty(true);
boolean containsJJ = false;
containsJJ = treeUtils.pocContainsJJs(poc);
// logger.info("Checking whether it contains JJ*..." + containsJJ);
if (containsJJ) {
if (alreadyAdded) clonedElement.setAlreadyAdded(true);
else clonedElement.setAlreadyAdded(false);
clonedElement.setAnnotation(element.getAnnotation());
clonedElement.setData(element.getData());
if (((PropertyElement) element).getRange() != null)
clonedElement.setRange(((PropertyElement) element).getRange());
if (((PropertyElement) element).getDomain() != null)
clonedElement.setDomain(((PropertyElement) element).getDomain());
if (((PropertyElement) element).getGovernor() != null)
clonedElement.setGovernor(((PropertyElement) element).getGovernor());
Vote vote = generateVote(text, propertyUri, clonedElement, FreyaConstants.MAX_FUNCTION);
votes.add(vote);
vote = generateVote(text, propertyUri, clonedElement, FreyaConstants.MIN_FUNCTION);
votes.add(vote);
vote = generateVote(text, propertyUri, clonedElement, FreyaConstants.SUM_FUNCTION);
votes.add(vote);
vote = generateVote(text, propertyUri, clonedElement, FreyaConstants.AVG_FUNCTION);
votes.add(vote);
}
}
}
return votes;
}
/**
* generate property elements from the set of property uris, and check whether these properties
* are datatype so that additional flag can be added to the property elements
*
* @param classes
* @return
*/
List<OntologyElement> returnPropertyElements(Set<String> properties) throws Exception {
if (properties == null) return new ArrayList();
List<OntologyElement> elements = new ArrayList<OntologyElement>();
for (String euri : properties) {
// logger.info("Checking if datatypeproperty=true for: " + euri);
SerializableURI elementUri = null;
try {
elementUri = new SerializableURI(euri, false);
} catch (Exception e) {
e.printStackTrace();
}
OntologyElement e = new PropertyElement();
// so that later we can distinguish property elements if they refer
// to datatypes
if (luceneAnnotator.isItDatatypeProperty(euri)) {
((PropertyElement) e).setDatatypeProperty(true);
}
e.setData(elementUri);
elements.add(((PropertyElement) e));
}
return elements;
}
/**
* checks whether the element is property and adds governor if yes
*
* @param pElement
* @param governor
* @return
*/
OntologyElement addGovernor(OntologyElement pElement, OntologyElement governor) {
PropertyElement property = null;
if (pElement instanceof PropertyElement) {
property = (PropertyElement) pElement;
// logger.debug("Property found:" + property.getData().toString()
// + "...is it datatype?" + property.isDatatypeProperty());
if (property.isDatatypeProperty()) {
// set governor for that element to be the
// neighbour.getOntologyElement()
property.setGovernor(governor);
// logger.debug("Setting governor:" + governor.toString()
// + " for element: " + property.getData().toString());
String uri = ((SerializableURI) property.getData()).toString();
Set<String> range = luceneAnnotator.findPropertyRange(uri);
String firstOne = null;
if (range != null && range.size() > 0) {
firstOne = (String) (new ArrayList(range)).get(0);
property.setRange(firstOne);
}
}
}
if (property == null) return pElement;
else return property;
}
/**
* find candidate ontology elements
*
* @return
*/
private List<OntologyElement> findCandidates(SemanticConcept sc, String text) throws Exception {
// Set<String> allSuggestions = new HashSet<String>();
OntologyElement el = sc.getOntologyElement();
// what is the population of california?
// what is the population of cities in california?
// what is the highest point of california?
// String uri = null;
Set<String> classUris = new HashSet<String>();
if (el instanceof InstanceElement) {
List<String> allClassUris = ((InstanceElement) el).getClassURIList();
logger.info("There were:" + allClassUris.size() + " direct types");
// classUris = filterDirectTypes(allClassUris);
classUris = new HashSet(allClassUris);
logger.info("There is NO FILTERING of direct types...");
} else if (el instanceof InstanceListElement) {
List allClassUris = ((InstanceListElement) el).getClassURIList();
logger.info("There were:" + allClassUris.size() + " direct types");
// classUris = filterDirectTypes(allClassUris);
classUris = new HashSet(allClassUris);
logger.info("There is NO FILTERING of direct types...");
} else if (el instanceof ClassElement) {
String uri = ((SerializableURI) el.getData()).toString();
classUris.add(uri);
logger.info("NN is class" + uri);
} else if (el instanceof DatatypePropertyValueElement) {
logger.info("NN is DPVE " + el.getData());
return findCandidatesForADTPV(((DatatypePropertyValueElement) el), text);
} else if (el instanceof PropertyElement) {
logger.info("NN is property " + el.getData());
return findCandidatesForAProperty(((PropertyElement) el), text);
}
Set<OntologyElement> elements = new HashSet<OntologyElement>();
Set<String> listOfPotentialCandidates = new HashSet<String>();
Set<String> classes = new HashSet<String>();
long start = System.currentTimeMillis();
for (String uri : classUris) {
// //////////properties first//////////////////////////
listOfPotentialCandidates = findCandidatesForClass(uri);
elements.addAll(returnPropertyElements(listOfPotentialCandidates));
// allSuggestions.addAll(listOfPotentialCandidates);
}
logger.info("Found " + elements.size() + " elements so far.");
// ////////////////////////////////////
for (String uri : classUris) {
classes.addAll(
luceneAnnotator.getNeighbouringClassesWhereGivenClassIsADomain(uri, forceSuperClasses));
logger.info("Found " + classes.size() + " class candidates from IsADomain method.");
classes.addAll(
luceneAnnotator.getNeighbouringClassesWhereGivenClassIsARange(uri, forceSuperClasses));
logger.info(
"Found more, now total: "
+ classes.size()
+ " class candidates from IsADomain and isARange method.");
}
Set<String> filteredClasses = new HashSet<String>();
for (String t : classes) {
if (!isInIgnoreNameSpaceList(t)) {
filteredClasses.add(t);
}
}
elements.addAll(returnClassElements(filteredClasses));
long end = System.currentTimeMillis();
logger.info(
"Found "
+ elements.size()
+ " cadidates for "
+ classUris.size()
+ " classes for "
+ (end - start)
+ "ms.");
return new ArrayList(elements);
}
/**
* @param property
* @param text
* @return
*/
private List<OntologyElement> findCandidatesForAProperty(PropertyElement property, String text)
throws Exception {
// Set<String> allSuggestions = new HashSet<String>();
String uri = ((SerializableURI) property.getData()).toString();
logger.debug("Finding candidates for property:" + property.getData().toString());
// all elements
List<OntologyElement> elements = new ArrayList<OntologyElement>();
// feed classes: these are used to find more suggestions but they are also
// added to the list!
Set<String> feedClasses = new HashSet<String>();
Set<String> properties = new HashSet<String>();
Set<String> allClasses = new HashSet<String>();
Set<String> rangeClasses = luceneAnnotator.findPropertyRange(uri);
feedClasses.addAll(rangeClasses);
logger.debug("getRangeClassesForProperty:" + uri + " found " + rangeClasses.size());
Set<String> domainClasses = luceneAnnotator.findPropertyDomain(uri);
feedClasses.addAll(domainClasses);
logger.debug("getDomainClassesForProperty:" + uri + " found " + domainClasses.size());
if (feedClasses == null || feedClasses.size() <= 1) {
// find top classes
logger.debug("Number of feed classes is 0....forceSuggestions=" + forceSuggestions);
if (new Boolean(forceSuggestions).booleanValue() == true) {
feedClasses.addAll(luceneAnnotator.findClassURIs());
}
}
// a onda za te classes nadji properties where class is a domain/range
for (String classUri : feedClasses) {
properties.addAll(
luceneAnnotator.getDefinedPropertiesWhereClassIsADomain(classUri, forceSuperClasses));
properties.addAll(
luceneAnnotator.getDefinedPropertiesWhereClassIsARange(classUri, forceSuperClasses));
}
//
if (new Boolean(forceSuggestions).booleanValue() == true) {
logger.info(
"cheating!!!!!!!!!!!!!!!!!!!!!!!!!!! here you should not get ALL properties but only relevant ones");
Set<String> datatypePropertiesList = luceneAnnotator.findDatatypePropertyURIs();
Set<String> objectPropertiesList = luceneAnnotator.findObjectPropertyURIs();
Set<String> rdfPropertiesList = luceneAnnotator.findRDFPropertyURIs(null);
properties.addAll(datatypePropertiesList);
properties.addAll(objectPropertiesList);
properties.addAll(rdfPropertiesList);
}
Set<String> filteredProperties = new HashSet<String>();
for (String t : properties) {
if (!isInIgnoreNameSpaceList(t)) {
filteredProperties.add(t);
}
}
// add datatype properties which do not have domain defined: mb endDate
// beginDate
// String table = getSparqlUtils().getDatatypePropertiesNoDomain();
// properties.addAll(gate.freya.util.StringUtil.fromStringToSet(table));
elements.addAll(returnPropertyElements(filteredProperties));
for (String classUri : feedClasses) {
allClasses.addAll(
luceneAnnotator.getNeighbouringClassesWhereGivenClassIsADomain(
classUri, forceSuperClasses));
allClasses.addAll(
luceneAnnotator.getNeighbouringClassesWhereGivenClassIsARange(
classUri, forceSuperClasses));
}
Set<String> filteredFeedClasses = new HashSet<String>();
for (String t : feedClasses) {
if (!isInIgnoreNameSpaceList(t)) {
filteredFeedClasses.add(t);
}
}
allClasses.addAll(filteredFeedClasses);
elements.addAll(returnClassElements(allClasses));
return elements;
}
