if DUMP_FEATURES:
trainFeaturesExtractor.DumpFeaturesToTxt(trainExportFileName)
testFeaturesExtractor.DumpFeaturesToTxt(testExportFileName)
if (CLASSIFIER):
# The serialization file to save the features
modelSerializationFile = ".\\Classifier\\Output\classifier_model.bin"
# Start the Classifier:
#---------------------
if (SVM_CLASSIFIER):
classifier = Classifier(modelSerializationFile, 'SVM',
trainFeaturesExtractor.features,
trainFeaturesExtractor.labels,
testFeaturesExtractor.features,
testFeaturesExtractor.labels)
if not LOAD_MODEL:
# Train
classifier.Train()
classifier.SaveModel()
else:
classifier.LoadModel()
# Test
labels, acc, val = classifier.Test()
# Build the confusion matrix
mConfusionMatrix, mNormalConfusionMatrix, vNumTrainExamplesPerClass, vAccuracyPerClass, nOverallAccuracy = classifier.BuildConfusionMatrix(
testFeaturesExtractor.labels, labels)
print(mConfusionMatrix)
classifier.SaveModel()
################## INDEPENDENT PROCESS TEST #################################
# Load language model from the the serialization file languageModelSerializationFile which was saved in training
testLanguageModel = LanguageModel(configFileLanguageModel, stopWordsFileName,
languageModelSerializationFile, linksDBFile,
datasetBuilder.trainSet)
testLanguageModel.LoadModel()
# Extract features
testFeaturesExtractor = FeaturesExtractor(configFileFeaturesExtractor,
testFeaturesSerializationFile,
testLabelsSerializationFile,
testLanguageModel,
datasetBuilder.testSet)
testFeaturesExtractor.ExtractNumTfFeatures()
# Load classifier
testClassifier = Classifier(configFileClassifier, modelSerializationFile,
trainFeaturesExtractor.features,
trainFeaturesExtractor.labels,
testFeaturesExtractor.features,
testFeaturesExtractor.labels)
testClassifier.LoadModel()
labels, acc, val = testClassifier.Test()
# Build the confusion matrix
mConfusionMatrix, mNormalConfusionMatrix, vNumTrainExamplesPerClass, vAccuracyPerClass, nOverallAccuracy = classifier.BuildConfusionMatrix(
testFeaturesExtractor.labels, labels)
print(mConfusionMatrix)