def testLoadLinksClassifier():
medMiningObj = MedGraphMining()
layerObj = NeuralLayerClassifier()
gensimModelPath = ROOT_PATH.auto_config_root(
) + u'model/word2vec/zongheword2vecModel.vector'
trainLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_train_links1-1200.txt'
testLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_test_links1201-1500.txt'
storeFilePath = ROOT_PATH.auto_config_root(
) + u'model/keras/links(sc2bz)_classifier_cnnlstmT'
layerModel = layerObj.loadStoredModel(storeFilePath, recompile=True)
# print(layerModel.to_json())
#===========================================================================
# classes, proba = medMiningObj.testLinksClasses_file(layerModel, gensimModelPath, testLinksDataPath)
#
# return classes, proba
#===========================================================================
score = medMiningObj.evaluateLinksClassifier_file(layerModel,
gensimModelPath,
testLinksDataPath)
print(score)
def testFindRelatBzFromPatient():
medMiningObj = MedGraphMining()
print('load w2v-model from path: ' + medMiningObj.medW2VModelPath)
ask = '胸闷，气喘，头昏四肢酸，嘴巴没味道，胸口胀气有时还一直打嗝，比以前瞌睡多，肺部没问题，只有点纵膈淋巴结肥大，心率64，去三甲医院看专家科医生说是哮喘，拿了哮喘药也没治到胸口气胀打嗝，还是头昏脚酸的，请问到底是肺部问题还是肠胃，还是心脏的问题？'
seg_ask = WordSeg('e').singlePosSegEngine(ask)
confBZRes = medMiningObj.findRelatBzFromPatient(seg_ask, 0.14)
for bz in confBZRes.keys():
print(bz + ' '),
print(confBZRes[bz])
return confBZRes
def testLoadLinksReps():
medMiningObj = MedGraphMining()
trainLinksFilePath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_train_links1-1000.txt'
textWordsList, maxTextLength, labelList = medMiningObj.loadLinksReps(
trainLinksFilePath)
for textWords in textWordsList:
print(' '.join(textWords))
print(len(textWordsList))
print(maxTextLength)
print(labelList)
def testSaveLinksClassifier():
medMiningObj = MedGraphMining()
layerObj = NeuralLayerClassifier()
gensimModelPath = ROOT_PATH.auto_config_root(
) + u'model/word2vec/zongheword2vecModel.vector'
trainLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_train_links1-1200.txt'
testLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_test_links1201-1500.txt'
storeFilePath = ROOT_PATH.auto_config_root(
) + u'model/keras/links(sc2bz)_classifier_cnnlstmT'
medMiningObj.trainLinksClassifier_file(gensimModelPath,
trainLinksDataPath,
v_ratio=0.15,
storeFilePath=storeFilePath)
def testEvaluateLinksClassify():
medMiningObj = MedGraphMining()
gensimModelPath = ROOT_PATH.auto_config_root(
) + u'model/word2vec/zongheword2vecModel.vector'
trainLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_train_links1-1200.txt'
testLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_test_links1201-1500.txt'
layerModel = medMiningObj.trainLinksClassifier_file(gensimModelPath,
trainLinksDataPath,
v_ratio=0.15)
score = medMiningObj.evaluateLinksClassifier_file(layerModel,
gensimModelPath,
testLinksDataPath)
print(score)
def testPredictNewLinks():
medMiningObj = MedGraphMining()
layerObj = NeuralLayerClassifier()
gensimModelPath = ROOT_PATH.auto_config_root(
) + u'model/word2vec/zongheword2vecModel.vector'
trainLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_train_links1-1200.txt'
predictLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_predict_links1501-2834.txt'
storeFilePath = ROOT_PATH.auto_config_root(
) + u'model/keras/links(sc2bz)_classifier_cnnlstm'
layerModel = layerObj.loadStoredModel(storeFilePath, recompile=False)
classes, proba = medMiningObj.testLinksClasses_file(
layerModel, gensimModelPath, predictLinksDataPath)
return classes, proba
def testClassifyLinks():
medMiningObj = MedGraphMining()
gensimModelPath = ROOT_PATH.auto_config_root(
) + u'model/word2vec/zongheword2vecModel.vector'
trainLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_train_links1-1200.txt'
testLinksDataPath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_test_links1201-1500.txt'
layerModel = medMiningObj.trainLinksClassifier_file(
gensimModelPath, trainLinksDataPath)
classes, proba = medMiningObj.testLinksClasses_file(
layerModel, gensimModelPath, testLinksDataPath)
# for i in range(len(classes)):
# print(str(classes[i]) + ': ' + str(proba[i]))
return classes, proba
def testLinksRepsToEmbeddingData():
medMiningObj = MedGraphMining()
layerObj = NeuralLayerClassifier()
load_start = time.clock()
trainLinksFilePath = ROOT_PATH.auto_config_root(
) + u'model_cache/relation_learning/shicai2bingzheng_train_links1-1200.txt'
textWordsList, maxTextLength, labelList = medMiningObj.loadLinksReps(
trainLinksFilePath)
gensimModelPath = ROOT_PATH.auto_config_root(
) + u'model/word2vec/zongheword2vecModel.vector'
x_data, y_data = layerObj.preTextEmbeddingProcess(gensimModelPath,
textWordsList,
maxTextLength, labelList)
load_end = time.clock()
print('load data run time: %f s' % (load_end - load_start))
print('x_data shape: ', x_data.shape)
print(x_data)
print('y_data shape: ', y_data.shape)
print(y_data)