def cosSim(self, doc1,doc2):
sim = 0
for k in doc1:
#if k in doc2:
a = (1 + math.log(doc1[k]))
b = (1+math.log(doc2[k]))
sim += a * b
if sim > 0:
doc1s = [1+math.log(doc1[k]) for k in doc1]
doc2s = [1+math.log(doc2[k]) for k in doc2]
sim = float(sim)/(getScalar(doc1s) * getScalar(doc2s))
else:
sim = 0
return sim
def __init__(self, topVocabDic,relevTh,docs):
self.docs = docs
self.relevanceth = relevTh
self.topVocabDic = topVocabDic
#doc1s = [1+math.log(self.topVocabDic[k]) for k in self.topVocabDic]
doc1s = [self.topVocabDic[k] for k in self.topVocabDic]
self.vocabScalar = getScalar(doc1s)
def cosSim(self,doc2):
sim = 0
#for k in doc1:
for k in self.topVocabDic:
if k in doc2:
#a = (1 + math.log(self.topVocabDic[k]))
#b = (1+math.log(doc2[k]))
a = self.topVocabDic[k]
b = doc2[k]
sim += a * b
if sim > 0:
#doc1s = [1+math.log(doc1[k]) for k in doc1]
#doc2s = [1+math.log(doc2[k]) for k in doc2]
doc2s = [doc2[k] for k in doc2]
#sim = float(sim)/(getScalar(doc1s) * getScalar(doc2s))
sim = float(sim)/(self.vocabScalar * getScalar(doc2s))
else:
sim = 0
return sim