def loadPairs(queryId, fileName, labeledPoints, pairLabels):
for line in open(fileName, 'r'):
split = line.split('\t')
cpoints = set()
for entry in split:
normQuery = filterWords(entry.strip())
#normQuery = entry.strip()#normalize(entry,stemmer)
if normQuery in queryId:
qid = queryId[normQuery]
labeledPoints.add(qid)
cpoints.add(qid)
#else:
#if len(normQuery) > 0:
# print normQuery
# pass
#print 'True same ',normQuery
pairs = generatePairsFromList(sorted(cpoints))
for pair in pairs:
pairLabels.add(pair)
def calculateIndiciesFromFiles(trueLabelFile, differentPairFile, predictedLabelFile,
queryList):
queryId = {}
idQuery = {}
i = 1
for line in open(queryList, 'r'):
split = line.split('\t')
typeList = ast.literal_eval(split[7])
if len(typeList) > 0:
query = filterWords(split[0].strip())
if query not in queryId:
queryId[query] = str(i)
idQuery[str(i)] = query
i += 1
l_samePairs = set()
l_points = set()
l_diffPairs = set()
p_samePairs = set()
#load true label file
#Same cluster
loadPairs(queryId, trueLabelFile, l_points, l_samePairs)
loadPairs(queryId, differentPairFile, l_points, l_diffPairs)
total_pairs = len(l_samePairs) + len(l_diffPairs) #(len(l_points)*(len(l_points)-1))/2
#filter predicted label file
for line in open(predictedLabelFile, 'r'):
line = line.strip()
cpoints = set()
if len(line) > 0 and 'NO CLUST' not in line:
split = line.split('\t')
for entry in split:
try:
entry = filterWords(entry.strip())
qid = queryId[entry]
if qid in l_points:
cpoints.add(qid)
except:
#print entry
pass
pairs = generatePairsFromList(sorted(cpoints))
for pair in pairs:
if pair in l_samePairs or pair in l_diffPairs:
#if pair in l_diffPairs:
# s = pair.split()
# print pair, idQuery[s[0]], idQuery[s[1]]
p_samePairs.add(pair) #
if pair in l_diffPairs:
p1 = pair[0:pair.find(' ')]
p2 = pair[pair.find(' ') + 1:]
#if p1 in idQuery and p2 in idQuery:
# print idQuery[p1], idQuery[p2]
#else:
# print p1, 'sec', p2
print len(l_samePairs), len(p_samePairs), total_pairs
#print l_samePairs
tp = len(l_samePairs & p_samePairs)
fp = len(p_samePairs & l_diffPairs) #len(p_samePairs) - tp ;
fn = len(l_samePairs) - tp
tn = len(l_diffPairs) - fp #total_pairs - (tp+fp+fn)
print tp, fp, fn, tn, total_pairs
return tp, fp, fn, tn, total_pairs