def test_Accuracy(self):
correct = 0
total = 0
lesk = Lesk(self.wordNet, self.fsm)
treeBank1 = TreeBankDrawable("../../new-trees")
treeBank2 = TreeBankDrawable("../../old-trees")
for i in range(treeBank1.size()):
parseTree1 = treeBank1.get(i)
parseTree2 = treeBank2.get(i)
lesk.autoSemantic(parseTree1)
nodeDrawableCollector1 = NodeDrawableCollector(
parseTree1.getRoot(), IsTurkishLeafNode())
leafList1 = nodeDrawableCollector1.collect()
nodeDrawableCollector2 = NodeDrawableCollector(
parseTree2.getRoot(), IsTurkishLeafNode())
leafList2 = nodeDrawableCollector2.collect()
for j in range(len(leafList1)):
total = total + 1
parseNode1 = leafList1[j]
parseNode2 = leafList2[j]
if parseNode1.getLayerData(
ViewLayerType.SEMANTICS
) is not None and parseNode1.getLayerData(
ViewLayerType.SEMANTICS) == parseNode2.getLayerData(
ViewLayerType.SEMANTICS):
correct = correct + 1
self.assertEqual(475, total)
self.assertEqual(247, correct)
def autoLabelSingleSemantics(self, parseTree: ParseTreeDrawable) -> bool:
random.seed(1)
nodeDrawableCollector = NodeDrawableCollector(parseTree.getRoot(),
IsTurkishLeafNode())
leafList = nodeDrawableCollector.collect()
done = False
for i in range(len(leafList)):
synSets = self.getCandidateSynSets(self.__turkishWordNet,
self.__fsm, leafList, i)
maxIntersection = -1
for j in range(len(synSets)):
synSet = synSets[j]
intersectionCount = self.intersection(synSet, leafList)
if intersectionCount > maxIntersection:
maxIntersection = intersectionCount
maxSynSets = []
for j in range(len(synSets)):
synSet = synSets[j]
if self.intersection(synSet, leafList) == maxIntersection:
maxSynSets.append(synSet)
if len(maxSynSets) > 0:
leafList[i].getLayerInfo().setLayerData(
ViewLayerType.SEMANTICS,
maxSynSets[randrange(len(maxSynSets))].getId())
done = True
return done
def autoLabelSingleSemantics(self, parseTree: ParseTreeDrawable) -> bool:
random.seed(1)
nodeDrawableCollector = NodeDrawableCollector(parseTree.getRoot(),
IsTurkishLeafNode())
leafList = nodeDrawableCollector.collect()
for i in range(len(leafList)):
synSets = self.getCandidateSynSets(self.__turkishWordNet,
self.__fsm, leafList, i)
if len(synSets) > 0:
leafList[i].getLayerInfo().setLayerData(
ViewLayerType.SEMANTICS,
synSets[randrange(len(synSets))].getId())
return True
def generateAnnotatedSentence(self,
language: str = None) -> AnnotatedSentence:
sentence = AnnotatedSentence()
if language is None:
nodeDrawableCollector = NodeDrawableCollector(
self.root, IsTurkishLeafNode())
leafList = nodeDrawableCollector.collect()
for parseNode in leafList:
if isinstance(parseNode, ParseNodeDrawable):
layers = parseNode.getLayerInfo()
for i in range(layers.getNumberOfWords()):
sentence.addWord(layers.toAnnotatedWord(i))
else:
nodeDrawableCollector = NodeDrawableCollector(
self.root, IsEnglishLeafNode())
leafList = nodeDrawableCollector.collect()
for parseNode in leafList:
if isinstance(parseNode, ParseNodeDrawable):
newWord = AnnotatedWord(
"{" + language + "=" + parseNode.getData().getName() +
"}{posTag=" +
parseNode.getParent().getData().getName() + "}")
sentence.addWord(newWord)
return sentence
def autoLabelSingleSemantics(self, parseTree: ParseTreeDrawable) -> bool:
nodeDrawableCollector = NodeDrawableCollector(parseTree.getRoot(),
IsTurkishLeafNode())
leafList = nodeDrawableCollector.collect()
for i in range(len(leafList)):
synSets = self.getCandidateSynSets(self.__turkishWordNet,
self.__fsm, leafList, i)
if len(synSets) > 0:
best = self.mostFrequent(
synSets,
leafList[i].getLayerInfo().getMorphologicalParseAt(
0).getWord().getName())
if best is not None:
leafList[i].getLayerInfo().setLayerData(
ViewLayerType.SEMANTICS, best.getId())
return True
def autoNER(self, parseTree: ParseTreeDrawable):
self.autoDetectPerson(parseTree)
self.autoDetectLocation(parseTree)
self.autoDetectOrganization(parseTree)
self.autoDetectMoney(parseTree)
self.autoDetectTime(parseTree)
nodeDrawableCollector = NodeDrawableCollector(
parseTree.getRoot(), IsTransferable(self.secondLanguage))
leafList = nodeDrawableCollector.collect()
for parseNode in leafList:
if isinstance(parseNode,
ParseNodeDrawable) and not parseNode.layerExists(
ViewLayerType.NER):
parseNode.getLayerInfo().setLayerData(ViewLayerType.NER,
"NONE")
parseTree.save()
def autoArgument(self, parseTree: ParseTreeDrawable, frameset: Frameset):
nodeDrawableCollector = NodeDrawableCollector(
parseTree.getRoot(), IsTransferable(self.secondLanguage))
leafList = nodeDrawableCollector.collect()
for parseNode in leafList:
if isinstance(parseNode,
ParseNodeDrawable) and parseNode.getLayerData(
ViewLayerType.PROPBANK) is None:
for argumentType in ArgumentType:
if frameset.containsArgument(
argumentType) and self.autoDetectArgument(
parseNode, argumentType):
parseNode.getLayerInfo().setLayerData(
ViewLayerType.PROPBANK,
ArgumentType.getPropbankType(argumentType))
if Word.isPunctuationSymbol(
parseNode.getLayerData(self.secondLanguage)):
parseNode.getLayerInfo().setLayerData(
ViewLayerType.PROPBANK, "NONE")
parseTree.save()
def extractNodesWithPredicateVerbs(self, wordNet: WordNet) -> list:
nodeDrawableCollector = NodeDrawableCollector(
self.root, IsPredicateVerbNode(wordNet))
return nodeDrawableCollector.collect()