def Training(args = sys.argv[2:]):
"""
Training RDRPOSTagger using initialized corpus against golden corpus!
The initialized corpus is already generated by the use of an external initial tagger.
"""
from src.tagger.SCRDRlearner.PosTaggingRDRTree import PosTaggingRDRTree
dirPath = os.path.join(args[0] + "/")
correctTrain = args[1]
initializedTrain = args[2]
learntRules = args[3]
print ('\nTraining RDRPOSTagger in the use of initialized corpus against golden corpus....')
print ('Building SCRDR-based POS tagging tree of rules...')
for (improveThreshold, matchThreshold) in thresholds:
outputDir = "T%d-%d/" % (improveThreshold, matchThreshold)
os.mkdir(dirPath + outputDir)
rdrTree = PosTaggingRDRTree(improveThreshold, matchThreshold)
rdrTree.buildTreeFromCorpus(dirPath + initializedTrain, dirPath + correctTrain)
print ("Write the tree to file...")
rdrTree.writeToFileWithoutSeenCases(dirPath + outputDir + learntRules)
#rdrTree.writeToFile(dirPath + outputDir + learntRules)
print ('Completed!')
def VnTraining(args = sys.argv[2:]):
pathToDict = args[0]
dirPath = os.path.join(args[1] + "/")
correctTrain = args[2]
learntRules = args[3]
print( '\nTraining RDRPOSTagger for Vietnamese POS Tagging...')
print( "Initial tagging...")
getRawTextFromFile(dirPath + correctTrain, dirPath + correctTrain + ".RAW")
DICT = readDictionary(pathToDict)
VnInitTagger4Corpus(DICT, dirPath + correctTrain + ".RAW", dirPath + correctTrain + ".INIT")
print ("Done Initialization!")
print ('Building SCRDR-based POS tagging tree of rules...')
for (improveThreshold, matchThreshold) in thresholds:
timeStart = time.time()
outputDir = "T%d-%d/" % (improveThreshold, matchThreshold)
os.mkdir(dirPath + outputDir)
rdrTree = PosTaggingRDRTree(improveThreshold, matchThreshold)
rdrTree.buildTreeFromCorpus(dirPath + correctTrain + ".INIT", dirPath + correctTrain)
print ("Write the tree to file...")
rdrTree.writeToFileWithoutSeenCases(dirPath + outputDir + learntRules)
#rdrTree.writeToFile(dirPath + outputDir + learntRules)
print ("\nTraining time for threshold %d-%d: %f seconds\n" % (improveThreshold, matchThreshold, time.time() - timeStart))
print ('\nCompleted!')
def TaggingInitializedCorpus(args = sys.argv[2:]):
learntRulesPath = args[0]
initTestCorpusPath = args[1]
print ("\nTagging initialized corpus:", initTestCorpusPath)
from src.tagger.SCRDRlearner.PosTaggingRDRTree import PosTaggingRDRTree
r = PosTaggingRDRTree()
r.constructTreeFromRulesFile(learntRulesPath)
r.tagInitializedCorpus_new(initTestCorpusPath, initTestCorpusPath + ".TAGGED")
print ('Completed!')
