refs = []
doneRefs = {}
for mapEntry, prediction in izip(mapFile, predictionFile):
referee, reference = mapEntry.strip().split(":")
label, confidence, invConfidence = prediction.strip().split()
if referee != currentReferee:
if currentReferee:
predictionOut = [x[0] for x in refs]
resultsFile.write(currentReferee + ", " +
" ".join(predictionOut) + "\n")
doneRefs[currentReferee] = 1
refs = []
currentReferee = referee
else:
if label == "1":
if reference in refCounts:
confidence = refCounts[reference]
refs = papers.appendMax((reference, confidence), refs, 10)
predictionOut = [x[0] for x in refs]
resultsFile.write(currentReferee + ", " + " ".join(predictionOut) + "\n")
doneRefs[currentReferee] = 1
with open(fullIdsLocation, "r") as fullIds:
for line in fullIds:
evalId, preds = line.split(",")
if evalId not in doneRefs:
resultsFile.write(evalId + ", " + "\n")
def testMaxAppendEnd(self):
test = [(1,3),(2,5),(3,8)]
res = papers.appendMax((4,9), test, 6)
self.assertEqual(res, [(1,3), (2,5), (3,8), (4,9)])
def testMaxAppendEmpty(self):
test = []
res = papers.appendMax((1,2), test, 3)
self.assertEqual(res, [(1,2)])
def testMaxAppendLess(self):
test = [(1,3),(2,5),(3,8)]
res = papers.appendMax((4,6), test, 4)
self.assertEqual(res, [(1,3), (2,5), (4,6), (3,8)])
print "----------------------------------------------------------------------------"
#heurisitic based prediction
predictions = []
with open(predictionLocation, "w", 0) as file:
file.write("Id,References\n")
for paper in validationData.papersByRef:
currentPaper = validationData.papersByRef[paper]
print "-Processing paper: ", paper
refs = []
if currentPaper.canonicalVenue in trainingData.venueReferences:
print "For paper: ", paper, " there are ", len(trainingData.venueReferences[currentPaper.canonicalVenue]), " venues"
for venue in trainingData.venueReferences[currentPaper.canonicalVenue]:
for nextPaperId in trainingData.indicesByCanonicalVenue[venue]:
nextPaper = trainingData.papersByRef[nextPaperId]
if(nextPaper.year < currentPaper.year and nextPaper.index != currentPaper.index):
sim = currentPaper.abstractCosineSimilarity(nextPaper.abstractList)
refs = papers.appendMax((nextPaper.index, sim), refs, 10)
prediction = [x[0] for x in refs]
# predictions.append((paper, ))
else:
prediction = []
# predictions.append((paper, []))
file.write(paper + ", " + " ".join(prediction) + "\n")
print "----------------------------------------------------------------------------"
print "- Number of training papers: ", len(trainingData.papersByRef)
print "- Number of validation papers: ", len(validationData.papersByRef)
print "----------------------------------------------------------------------------"
def testMaxAppendEmpty(self):
test = []
res = papers.appendMax((1, 2), test, 3)
self.assertEqual(res, [(1, 2)])
def testMaxAppendEnd(self):
test = [(1, 3), (2, 5), (3, 8)]
res = papers.appendMax((4, 9), test, 6)
self.assertEqual(res, [(1, 3), (2, 5), (3, 8), (4, 9)])
def testMaxAppendLess(self):
test = [(1, 3), (2, 5), (3, 8)]
res = papers.appendMax((4, 6), test, 4)
self.assertEqual(res, [(1, 3), (2, 5), (4, 6), (3, 8)])