def getTrees(self):
self.data.setCurrentId(23)
#for item in self.data.getTextTrees():
#for itemtwo in self.data.getHypTrees():
item = self.data.getTextTree()
itemtwo = self.data.getHypTree()
print "ONE ", item, "\nTWO ", itemtwo
dist = td.distance(item, itemtwo, weighter)
empty = td.Node("TEST")
norm = td.distance(empty, itemtwo, weighter)
print "distance ", dist, "normalisation ", norm, " fin ", (norm-dist)/norm
def matchTrees(self, weight = False ):
def weighter(node1, node2):
# insertion cost
if node1 is None:
if node2.label in ["ABSTRACT", "ROOT"]:
return 0
#print "label ", node2.label, tdi.data.getDf(node2.label, "lemma")
return self.data.getDf(node2.label, "lemma")
# deletion cost
if node2 is None:
return 0
# substitution cost
if node1.label != node2.label:
#if node1.label in ["ABSTRACT", "ROOT"] and node2.label in ["ABSTRACT", "ROOT"]:
# return 0
#return (tdi.data.getDf(node1.label, "lemma")+tdi.data.getDf(node2.label, "lemma"))/2
return 1
return 0
#PART IV
def weighterSyn(node1, node2):
# insertion cost
if node1 is None:
if node2.label in ["ABSTRACT", "ROOT"]:
return 0
#print "label ", node2.label, tdi.data.getDf(node2.label, "lemma")
return self.data.getDf(node2.label, "lemma")
# deletion cost
if node2 is None:
return 0
# substitution cost
if node1.label != node2.label:
#if node1.label in ["ABSTRACT", "ROOT"] and node2.label in ["ABSTRACT", "ROOT"]:
# return 0
#return (tdi.data.getDf(node1.label, "lemma")+tdi.data.getDf(node2.label, "lemma"))/2
node1Syn = syn.Syn(node1.label)
node2Syn = syn.Syn(node2.label)
return 1-node1Syn.findLemmaConnection(node2Syn)
return 0
for i in range(1, self.data.pairs+1):
if i % 8 == 0:
print i*100/800,"%"
self.data.setCurrentId(i)
tTree = self.data.getTextTree()
hTree = self.data.getHypTree()
empty = td.Node("EMPTY")
if weight:
dist = td.distance(tTree, hTree, weighterSyn)
norm = td.distance(empty, hTree, weighterSyn)
else:
dist = td.distance(tTree, hTree)
norm = td.distance(empty, hTree)
#self.result[i] = "YES" if (norm-dist)/norm > self.thresh else "NO"
self.result[i] = (norm-dist)/norm
return self.result
