def pattern_tokenText_bussvc(token, text, command):
list_svc = load_dict('svc')
list_patt = pattern_bus_service_ver2(text, list_svc, command)
if token in list_patt:
return True
else:
return False
def check_reg(svc, text, list_sv):
# if the text contain correct bus service which appear in text, add word "feature_reg"
list_pattern_services = pattern_bus_service_ver2(text, list_sv)
list_match_services = pattern_bus_service(text, list_sv)
list_total = list(set(list_pattern_services) | set(list_match_services))
if svc in list_total:
return text + ' feature_reg'
else:
return text
def extract_svc_expression(list_line, list_sv):
y_label = []
y_reg = []
list_svc = []
for line in list_line:
split_line = line.split('\t')
index = split_line[0]
label = split_line[1]
svc = split_line[2].strip() # note that svc can be string or int => format svc as string
list_svc.append(svc)
text = split_line[3].strip()
# print index, label, svc
y_label.append(split_line[1])
list_pattern_services = pattern_bus_service_ver2(text, list_sv)
list_match_services = pattern_bus_service(text, list_sv)
list_total = list(set(list_pattern_services) | set(list_match_services))
if svc in list_total:
y_reg.append('TRUE')
else:
y_reg.append('FALSE')
for value in y_reg:
print value
# for i in range(0, len(y_reg)):
# if y_label[i] != y_reg[i]:
# print list_svc[i]
print metrics.accuracy_score(y_label, y_reg)
print metrics.classification_report(y_label, y_reg)
print metrics.confusion_matrix(y_label, y_reg)