def aggregateResults(results):
"""Aggregates results from crosswikis by ignoring case on the anchor.
Numerators are added up, and probabilities are averaged, weighted by their
numerators.
Args:
results: a results set of the form [(anchor, entity, cprob, num, denom)]
Returns: a new results set of the form [(anchor, entity, cprob, num, denom)]
"""
linkCounts = {}
linkCprobs = {}
for anchor, entity, info, cprob in results:
anchor=anchor.lower()
labelCounts = getLabelCounts(info)
num = sum([num for (num, denom) in labelCounts.values()])
myutils.addToDict(linkCounts, (anchor, entity), num)
myutils.addToDict(linkCprobs, (anchor, entity), num*cprob)
denom = sum(linkCounts.values())
results = []
for ((anchor, entity), num) in linkCounts.items():
cprob = linkCprobs[(anchor, entity)] / denom
results.append((anchor, entity, cprob, num, denom))
return results
def getNumInstances(instancesFile):
instanceCounts = {}
for line in instancesFile:
lineParts = line.strip().split('\t')
string = lineParts[0]
count = int(lineParts[1])
myutils.addToDict(instanceCounts, string, count)
return instanceCounts
def getNumInstances(instancesFile):
instanceCounts = {}
for line in instancesFile:
lineParts = line.strip().split('\t')
string = lineParts[0]
count = int(lineParts[1])
myutils.addToDict(instanceCounts, string, count)
return instanceCounts
def dedupePr():
prFile = open(PR_OUTPUT_PATH)
dedupedPrFile = open(PR_DEDUPED_PATH, 'w')
prResults = readPrFile(prFile)
prCounts = {}
prParams = {}
for p, c, t, precision, recall in prResults:
myutils.addToDict(prCounts, (precision, recall), 1)
prParams[(precision, recall)] = 'p={}, c={}, t={}'.format(p, c, t)
for ((precision, recall), count) in prCounts.items():
paramString = prParams[(precision, recall)]
print('{}\t{}\t{}\t{}'.format(paramString, precision, recall, count),
file=dedupedPrFile)
def dedupePr():
prFile = open(PR_OUTPUT_PATH)
dedupedPrFile = open(PR_DEDUPED_PATH, 'w')
prResults = readPrFile(prFile)
prCounts = {}
prParams = {}
for p, c, t, precision, recall in prResults:
myutils.addToDict(prCounts, (precision, recall), 1)
prParams[(precision, recall)] = 'p={}, c={}, t={}'.format(p, c, t)
for ((precision, recall), count) in prCounts.items():
paramString = prParams[(precision, recall)]
print('{}\t{}\t{}\t{}'.format(
paramString,
precision,
recall,
count
),
file=dedupedPrFile
)
