def sent_sim_score_num_nn_words(sent1,sent2):
words1 = sent_to_words(sent1.lower())
words2 = sent_to_words(sent2.lower())
denom = math.log(len(set(words1))) + math.log(len(set(words2)))
if denom < 2:
return 0
pos_tags1 = fn_assign_POS_tags(words1)
pos_tags2 = fn_assign_POS_tags(words2)
num_meaningful = 0
for counter1 in range(len(words1)):
for counter2 in range(len(words2)):
if words1[counter1]==words2[counter2]:
if pos_tags1[counter1] in ['ADJ','N','NP'] or pos_tags2[counter2] in ['ADJ','N','NP']:
num_meaningful+=1
return num_meaningful/denom
def enhanced_rouge(n,reference_sents,machine_sents,counter):
ref_n_grams = set()
for se in reference_sents:
words = sent_to_words(se)
for index in range(len(words)+1-n):
ref_n_grams.add(' '.join(words[index:index+n]))
if len(ref_n_grams)==0:
return -1
mac_n_grams = set()
for se in machine_sents:
#print "se=",se
words = sent_to_words(se)
for index in range(len(words)+1-n):
mac_n_grams.add(' '.join(words[index:index+n]))
special = set()
intersect = ref_n_grams&mac_n_grams
if n > 1:
uniq_mac = mac_n_grams - ref_n_grams
for uniq in uniq_mac:
ws = uniq.split()
pos_tags = fn_assign_POS_tags(ws)
for index in range(len(ws)):
if pos_tags[index] in ['ADJ','N','NP'] :
for ref in ref_n_grams:
if ref.find(ws[index])!=-1:
special.add(uniq)
num = len(special)/n
return (len(intersect)+num)/len(ref_n_grams)
