args.num_classes = 2
args.input_dim = 300
args.sparse = ""
args.lr = 0.01
args.wd = 1e-4
args.data = os.path.join(base_path, "data/lc_quad/")
args.cuda = False
parser = LC_QaudParser()
kb = parser.kb
base_dir = "./output"
question_type_classifier_path = os.path.join(base_dir,
"question_type_classifier")
utility.makedirs(question_type_classifier_path)
question_type_classifier = SVMClassifier(
os.path.join(question_type_classifier_path, "svm.model"))
o = Orchestrator(None, question_type_classifier, None, parser, True)
raw_entities = [{
"surface":
"",
"uris": [{
"confidence": 1,
"uri": "http://dbpedia.org/resource/Bill_Finger"
}]
}]
entities = []
for item in raw_entities:
uris = [
Uri(uri["uri"], kb.parse_uri, uri["confidence"])
for uri in item["uris"]
ds.load()
ds.parse()
if not ds.parser.kb.server_available:
logger.error("Server is not available. Please check the endpoint at: {}".format(ds.parser.kb.endpoint))
sys.exit(0)
output_file = 'lcquadtestanswer_output'
linker = Earl(path="data/LC-QUAD/entity_lcquad_test.json")
base_dir = "./output"
question_type_classifier_path = os.path.join(base_dir, "question_type_classifier")
double_relation_classifier_path = os.path.join(base_dir, "double_relation_classifier")
utility.makedirs(question_type_classifier_path)
utility.makedirs(double_relation_classifier_path)
question_type_classifier = SVMClassifier(os.path.join(question_type_classifier_path, "svm.model"))
double_relation_classifier = SVMClassifier(os.path.join(double_relation_classifier_path, "svm.model"))
stats = Stats()
parser = LC_QaudParser()
kb = parser.kb
o = Orchestrator(logger, question_type_classifier, double_relation_classifier, parser, question_type_classifier_path, True)
tmp = []
output = []
na_list = []
for qapair in ds.qapairs:
stats.inc("total")
from parser.lc_quad import LC_QaudParser
from sklearn.metrics import classification_report
def run(classifier1, classifier2):
parser = LC_QaudParser()
query_builder = Orchestrator(classifier1,
classifier2,
parser,
auto_train=False)
print "train_question_classifier"
scores = query_builder.train_question_classifier(
file_path="../data/LC-QUAD/data_v8.json", test_size=0.5)
print scores
y_pred = query_builder.question_classifier.predict(query_builder.X_test)
print(classification_report(query_builder.y_test, y_pred))
print "double_relation_classifer"
scores = query_builder.train_double_relation_classifier(
file_path="../data/LC-QUAD/data_v8.json", test_size=0.5)
print scores
y_pred = query_builder.double_relation_classifer.predict(
query_builder.X_test)
print(classification_report(query_builder.y_test, y_pred))
if __name__ == "__main__":
run(SVMClassifier(), SVMClassifier())
run(NaiveBayesClassifier(), NaiveBayesClassifier())
query_builder = Orchestrator(None,
classifier1,
classifier2,
parser,
auto_train=False)
print("train_question_classifier")
scores = query_builder.train_question_classifier(
file_path="../data/LC-QUAD/data_v8.json", test_size=0.5)
print(scores)
y_pred = query_builder.question_classifier.predict(query_builder.X_test)
print(classification_report(query_builder.y_test, y_pred))
print("double_relation_classifer")
scores = query_builder.train_double_relation_classifier(
file_path="../data/LC-QUAD/data_v8.json", test_size=0.5)
print(scores)
y_pred = query_builder.double_relation_classifer.predict(
query_builder.X_test)
print(classification_report(query_builder.y_test, y_pred))
if __name__ == "__main__":
run(SVMClassifier('../output/question_type_classifier/svm.model'),
SVMClassifier('../output/double_relation_classifier/svm.model'))
run(
NaiveBayesClassifier(
'../output/question_type_classifier/naivebayes.model'),
NaiveBayesClassifier(
'../output/double_relation_classifier/naivebayes.model'))
from parser.lc_quad import LC_QaudParser
from learning.classifier.svmclassifier import SVMClassifier
from parser.qald import Qald
from parser.lc_quad_linked import LC_Qaud_Linked
from sklearn.metrics import classification_report
from sklearn.metrics import accuracy_score
import sys
import numpy as np
import json
from sklearn.metrics import confusion_matrix
import matplotlib.pyplot as plt
if __name__ == "__main__":
parser = LC_QaudParser()
classifier1 = SVMClassifier('./output/question_type_classifier/svm.model')
classifier2 = SVMClassifier('./output/double_relation_classifier/svm.model')
query_builder = Orchestrator(None, classifier1, classifier2, parser, None, auto_train=True)
print("train_question_classifier")
scores = query_builder.train_question_classifier(file_path="./data/LC-QUAD/data.json", test_size=0.8)
print(scores)
y_pred = query_builder.question_classifier.predict(query_builder.X_test)
print(accuracy_score(query_builder.y_test, y_pred))
print(classification_report(query_builder.y_test, y_pred, digits=3))
ds = LC_Qaud_Linked(path="./data/LC-QUAD/linked_test.json")
ds.load()
ds.parse()