def main():
print u.now(), "Start"
modelFile = "tag.model"
inputFile = "gene.dev"
outputFile = "dev.p1.out"
# inputFile = "short.dev"
# outputFile = "dev.short.out"
inputFile = "gene.test"
outputFile = "gene_test.p1.out"
weightVector = readPreTrainedModel(modelFile)
lines = u.readFromFile(inputFile)
tagSequences = []
while len(lines) > 0:
sentence, lines = u.GetNextSentence(lines)
featureData = FeatureData(weightVector)
tagSequence = ViterbiGLM(featureData, sentence)
tagSequences.append((tagSequence, sentence))
outputText = []
lines = u.readFromFile(inputFile)
del (tagSequences[0][0])[0]
del (tagSequences[0][1])[0]
for tagSequence, sentence in tagSequences:
for j in xrange(0,len(tagSequence)):
if len(tagSequence) != len(sentence):
continue
tag = tagSequence[j]
word = sentence[j]
outputText.append("%s %s" % (word, tag))
outputText.append("")
outputText.append("")
outputText = '\n'.join(outputText)
u.saveToFile(outputFile, outputText)
print u.now(), "Done"
def main():
print u.now(), "Start"
modelFile = "tag.model"
trainingFile = "gene.train"
inputFile = "gene.dev"
outputFile = "dev.p1.out"
lines = u.readFromFile(inputFile)
#weightVector = p1.readPreTrainedModel(modelFile)
# Empty weight vector means a vector of 0s
weightVector = dict()
featureData = FeatureData(weightVector)
trainingSet = readTrainingSetFile(trainingFile)
perceptron(trainingSet, featureData, 5)
# tagSequences = []
#
# while len(lines) > 0:
#sentence, lines = u.GetNextSentence(lines)
# genX = gen(["", "hello", "there", "foo", "bar", "bacon"])
print u.now(), "Done"
def readPreTrainedModel(modelFilename):
lines = u.readFromFile(modelFilename)
vMap = dict()
for i in xrange(0, len(lines)):
line = lines[i]
tokens = line.split()
if len(tokens) != 2:
print "Invalid line %s: %s" % ( i, line )
vMap[tokens[0]] = float(tokens[1])
return vMap
def readTrainingSetFile(filename):
lines = u.readFromFile(filename)
trainingSet = []
while len(lines) > 0:
trainingSentence, lines = u.GetNextSentence(lines)
x = []
y = []
for pair in trainingSentence[1:]:
splitPair = pair.split()
x.append(splitPair[0])
y.append(splitPair[1])
trainingSet.append ( (x,y) )
return trainingSet
