def computeSimilarity(self, synSet1: SynSet, synSet2: SynSet) -> float:
pathToRootOfSynSet1 = self.wordNet.findPathToRoot(synSet1)
pathToRootOfSynSet2 = self.wordNet.findPathToRoot(synSet2)
LCSid = self.wordNet.findLCSid(pathToRootOfSynSet1,
pathToRootOfSynSet2)
return 1 / (self.informationContents[synSet1.getId()] +
self.informationContents[synSet2.getId()] -
2 * self.informationContents[LCSid])
def __containsSameLiteral(self, synSet1: SynSet, synSet2: SynSet) -> bool:
for i in range(synSet1.getSynonym().literalSize()):
literal1 = synSet1.getSynonym().getLiteral(i)
for j in range(i + 1, synSet2.getSynonym().literalSize()):
literal2 = synSet2.getSynonym().getLiteral(j)
if literal1.getName() == literal2.getName() and synSet1.getPos(
) is not None:
return True
return False
def intersection(self, synSet: SynSet, sentence: AnnotatedSentence) -> int:
if synSet.getExample() is not None:
words1 = (synSet.getLongDefinition() + " " +
synSet.getExample()).split(" ")
else:
words1 = synSet.getLongDefinition().split(" ")
words2 = sentence.toString().split(" ")
count = 0
for word1 in words1:
for word2 in words2:
if word1.lower() == word2.lower():
count = count + 1
return count
def changeSynSetId(self, synSet: SynSet, newId: str):
"""
Changes ID of a specified SynSet with the specified new ID.
PARAMETERS
----------
synSet : SynSet
SynSet whose ID will be updated
newId : str
new ID
"""
self.__synSetList.pop(synSet.getId())
synSet.setId(newId)
self.__synSetList[newId] = synSet
def intersection(self, synSet: SynSet, leafList: list) -> int:
if synSet.getExample() is not None:
words1 = (synSet.getLongDefinition() + " " +
synSet.getExample()).split(" ")
else:
words1 = synSet.getLongDefinition().split(" ")
words2 = []
for i in range(len(leafList)):
words2.append(leafList[i].getLayerData(ViewLayerType.TURKISH_WORD))
count = 0
for word1 in words1:
for word2 in words2:
if word1.lower() == word2.lower():
count = count + 1
return count
def removeSynSet(self, synSet: SynSet):
"""
Removes specified SynSet from the SynSet list.
PARAMETERS
----------
synSet : SynSet
SynSet to be removed
"""
self.__synSetList.pop(synSet.getId())
def addSynSet(self, synSet: SynSet):
"""
Adds specified SynSet to the SynSet list.
PARAMETERS
----------
synSet : SynSet
SynSet to be added
"""
self.__synSetList[synSet.getId()] = synSet
def findPathToRoot(self, synSet: SynSet) -> list:
"""
Finds the path to the root node of a SynSets.
PARAMETERS
----------
synSet : SynSet
SynSet whose root path will be found
RETURNS
-------
list
List of String corresponding to nodes in the path
"""
pathToRoot = []
while synSet is not None:
if synSet.getId() in pathToRoot:
break
pathToRoot.append(synSet.getId())
synSet = self.percolateUp(synSet)
return pathToRoot
def percolateUp(self, root: SynSet) -> SynSet:
"""
Finds the parent of a node. It does not move until the root, instead it goes one level up.
PARAMETERS
----------
root : SynSet
SynSet whose parent will be find
RETURNS
-------
SynSet
Parent SynSet
"""
for i in range(root.relationSize()):
r = root.getRelation(i)
if isinstance(r, SemanticRelation):
if r.getRelationType() == SemanticRelationType.HYPERNYM \
or r.getRelationType() == SemanticRelationType.INSTANCE_HYPERNYM:
root = self.getSynSetWithId(r.getName())
return root
return None
def removeReverseRelation(self, synSet: SynSet,
semanticRelation: SemanticRelation):
"""
Removes the reverse relations from the SynSet.
PARAMETERS
----------
synSet : SynSet
SynSet to remove the reverse relation
semanticRelation : SemanticRelation
relation whose reverse will be removed
"""
otherSynSet = self.getSynSetWithId(semanticRelation.getName())
if otherSynSet is not None and SemanticRelation.reverse(
semanticRelation.getRelationType()) is not None:
otherRelation = SemanticRelation(
synSet.getId(),
SemanticRelation.reverse(semanticRelation.getRelationType()))
if otherSynSet.containsRelation(otherRelation):
otherSynSet.removeRelation(otherRelation)
def addReverseRelation(self, synSet: SynSet,
semanticRelation: SemanticRelation):
"""
Adds the reverse relations to the SynSet.
PARAMETERS
----------
synSet : SynSet
SynSet to add the reverse relations
semanticRelation : SemanticRelation
relation whose reverse will be added
"""
otherSynSet = self.getSynSetWithId(semanticRelation.getName())
if otherSynSet is not None and SemanticRelation.reverse(
semanticRelation.getRelationType()) is not None:
otherRelation = SemanticRelation(
synSet.getId(),
SemanticRelation.reverse(semanticRelation.getRelationType()))
if not otherSynSet.containsRelation(otherRelation):
otherSynSet.addRelation(otherRelation)
def __init__(self, fileName: str = None, exceptionFileName: str = None):
"""
Constructor that initializes the SynSet list, literal list, reads exception.
PARAMETERS
----------
fileName : str
Resource to be read for the WordNet.
"""
self.__exceptionList = {}
if fileName is None:
fileName = pkg_resources.resource_filename(
__name__, 'data/turkish_wordnet.xml')
elif exceptionFileName is not None:
self.readExceptionFile(exceptionFileName)
self.__interlingualList = {}
self.__synSetList = OrderedDict()
self.__literalList = OrderedDict()
root = xml.etree.ElementTree.parse(fileName).getroot()
currentSynSet = None
for synSetNode in root:
for partNode in synSetNode:
if partNode.tag == "ID":
currentSynSet = SynSet(partNode.text)
self.addSynSet(currentSynSet)
elif partNode.tag == "DEF":
currentSynSet.setDefinition(partNode.text)
elif partNode.tag == "EXAMPLE":
currentSynSet.setExample(partNode.text)
elif partNode.tag == "WIKI":
currentSynSet.setWikiPage(partNode.text)
elif partNode.tag == "BCS":
currentSynSet.setBcs(int(partNode.text))
elif partNode.tag == "POS":
if partNode.text == "a":
currentSynSet.setPos(Pos.ADJECTIVE)
elif partNode.text == "v":
currentSynSet.setPos(Pos.VERB)
elif partNode.text == "b":
currentSynSet.setPos(Pos.ADVERB)
elif partNode.text == "n":
currentSynSet.setPos(Pos.NOUN)
elif partNode.text == "i":
currentSynSet.setPos(Pos.INTERJECTION)
elif partNode.text == "c":
currentSynSet.setPos(Pos.CONJUNCTION)
elif partNode.text == "p":
currentSynSet.setPos(Pos.PREPOSITION)
elif partNode.text == "r":
currentSynSet.setPos(Pos.PRONOUN)
elif partNode.tag == "SR":
if len(partNode) > 0 and partNode[0].tag == "TYPE":
typeNode = partNode[0]
if len(partNode) > 1 and partNode[1].tag == "TO":
toNode = partNode[1]
currentSynSet.addRelation(
SemanticRelation(partNode.text, typeNode.text,
int(toNode.text)))
else:
currentSynSet.addRelation(
SemanticRelation(partNode.text, typeNode.text))
elif partNode.tag == "ILR":
if len(partNode) > 0 and partNode[0].tag == "TYPE":
typeNode = partNode[0]
interlingualId = partNode.text
if interlingualId in self.__interlingualList:
synSetList = self.__interlingualList[
interlingualId]
else:
synSetList = []
synSetList.append(currentSynSet)
self.__interlingualList[interlingualId] = synSetList
currentSynSet.addRelation(
InterlingualRelation(interlingualId,
typeNode.text))
elif partNode.tag == "SYNONYM":
for literalNode in partNode:
currentLiteral = None
for childNode in literalNode:
if childNode.tag == "SENSE":
currentLiteral = Literal(
literalNode.text, int(childNode.text),
currentSynSet.getId())
currentSynSet.addLiteral(currentLiteral)
self.addLiteralToLiteralList(currentLiteral)
elif childNode.tag == "SR":
typeNode = childNode[0]
if len(childNode
) > 1 and childNode[1].tag == "TO":
toNode = childNode[1]
currentLiteral.addRelation(
SemanticRelation(
childNode.text, typeNode.text,
int(toNode.text)))
else:
currentLiteral.addRelation(
SemanticRelation(
childNode.text, typeNode.text))