def download_data_and_compare_sentence(
hostname=["http://192.168.50.139:8081/api/"],
dirpath="/opt/lavector/sentence/",
jsonfile=["192.168.50.139_200_8085_0517.json"],
usecache=True):
"""
从label_studio的某个hostname下载数据，然后预测，最后给出结果
:return:
"""
from sentence_api import export_data
#从label-studio下载文
if usecache:
json_files = [os.path.join(dirpath, j) for j in jsonfile]
else:
json_files = []
for hname, jfile in zip(hostname, jsonfile):
json_file = export_data(hostname=hname,
dirpath=dirpath,
jsonfile=jfile,
proxy=False)
json_files.append(json_file)
original_data = []
for json_file in json_files:
#加载从label-studio获取的到json文件
with open(json_file, 'r') as f:
data = json.load(f)
print(f"共收集主机{json_file}的数据{len(data)} 条")
original_data.extend(data)
data = predict_comare_excel_sentence(original_data)
return data
def download_data_and_compare(hostname=["http://192.168.50.139:8081/api/"],
dirpath="/opt/lavector/absa/",
jsonfile=["192.168.50.139_200_8081_0517.json"],
service="absa",
usecache=True):
"""
从label_studio的某个hostname下载数据，然后预测，最后给出结果
:param service: "absa", "dem8", 使用哪个接口的服务进行预测
:return:
"""
from absa_api import export_data
#从label-studio下载文
if usecache:
json_files = [os.path.join(dirpath, j) for j in jsonfile]
else:
json_files = []
for hname, jfile in zip(hostname, jsonfile):
json_file = export_data(hostname=hname,
dirpath=dirpath,
jsonfile=jfile,
proxy=False)
json_files.append(json_file)
original_data = []
for json_file in json_files:
#加载从label-studio获取的到json文件
with open(json_file, 'r') as f:
data = json.load(f)
print(f"共收集主机{json_file}的数据{len(data)} 条")
original_data.extend(data)
# original_data = original_data[:600]
# data = predict_comare_excel(original_data, isabsa=isabsa)
data = predict_comare_excel_bad_case(original_data, service=service)
return data
def download_data_and_compare(hostname=["http://192.168.50.139:8081/api/"],
dirpath="/opt/lavector/absa/",
jsonfile=["192.168.50.139_500_8081_0129.json"],
isabsa=True,
result_excel="result.xlsx",
export_wrong_examples_excel="wrong.xlsx",
correct_examples_excel="correct.xlsx",
type=None):
"""
从label_studio的某个hostname下载数据，然后预测，最后给出结果
:return:
"""
from absa_api import export_data
#从label-studio下载文
original_data = []
for hname, jfile in zip(hostname, jsonfile):
json_file = export_data(hostname=hname,
dirpath=dirpath,
jsonfile=jfile,
proxy=False)
#加载从label-studio获取的到json文件
with open(json_file, 'r') as f:
data = json.load(f)
print(f"共收集主机{hname}的数据{len(data)} 条")
original_data.extend(data)
data = predict_comare_excel(
original_data,
result_excel=result_excel,
export_wrong_examples_excel=export_wrong_examples_excel,
correct_examples_excel=correct_examples_excel,
isabsa=isabsa,
type=type)
return data
def download_data_and_compare_same(hostname=[
"http://192.168.50.139:8081/api/", "http://192.168.50.139:8085/api/"
],
dirpath="/opt/lavector/absa/",
jsonfile=[
"192.168.50.139_500_8081_0129.json",
"192.168.50.139_500_8085_0129.json"
],
isabsa=True):
"""
对比相同的hostname的数据
从label_studio的某个hostname下载数据，然后预测，最后给出结果
:return:
"""
from absa_api import export_data
#从label-studio下载文
if len(hostname) != 2:
raise Exception("必须准2个hostname，里面包含相同的评估数据")
result = []
for hname, jfile in zip(hostname, jsonfile):
original_data = []
json_file = export_data(hostname=hname,
dirpath=dirpath,
jsonfile=jfile,
proxy=False)
#加载从label-studio获取的到json文件
with open(json_file, 'r') as f:
data = json.load(f)
print(f"共收集主机{hname}的数据{len(data)} 条")
original_data.extend(data)
predict_data, excel_data = predict_comare_excel(original_data,
isabsa=isabsa)
result.append([hname, predict_data, excel_data])
#对比2个人标注的数据
diffrent_data = []
print(f"对比host为 {result[0][0], result[1][0]}")
hname1, data1, pre1 = result[0]
hname2, data2, pre2 = result[1]
if len(data1) != len(data2):
raise Exception("两个人标注的数据总数不一致")
for d1, d2 in zip(data1, data2):
if d1[0] != d2[0]:
print("这条数据不一致")
else:
if d1[4] != d2[4]:
print(f"2个人标注的标签不一致")
print(d1[0])
print(d1[1])
print(d1[4])
print(d2[4])
one_data = {
"text": d1[0],
"keyword": d1[1],
"P1_label": d1[4],
"P2_label": d2[4],
"location": d1[2:4]
}
diffrent_data.append(one_data)
print(f"不一致的数据总量是{len(diffrent_data)}")
df = pd.DataFrame(diffrent_data)
writer = pd.ExcelWriter("diffrent.xlsx", engine='xlsxwriter')
df.to_excel(writer)
writer.save()
print(f"保存到diffrent.xlsx excel成功")
return data