def START_AUTO(filename):
read = MAR()
read = read.create(filename)
pos_last = 0
full_life = 3
life = full_life
while True:
pos, neg, total = read.get_numbers()
print("%d/ %d" % (pos, pos + neg))
if pos >= 10:
if pos == pos_last:
life = life - 1
if life == 0:
break
else:
life = full_life
if pos == 0:
for id in read.random():
read.code(id, read.body["label"][id])
else:
a, b, ids, c = read.train()
for id in ids:
read.code(id, read.body["label"][id])
pos_last = pos
return read
def START(filename, cl="linear"):
stop = 1
thres = 40
# thres=10000000000
read = MAR()
read = read.create(filename)
target = int(read.get_allbugs() * stop)
while True:
found, cost, total = read.get_numbers()
try:
print("%d, %d" % (found, cost))
except:
pass
if found >= target:
break
if found == 0 or cost < thres:
for id in read.loc_sort():
read.code(id, read.body["label"][id])
else:
ids, c = read.train(cl=cl)
for id in ids:
read.code(id, read.body["label"][id])
read.plot()
set_trace()
return read
def START_est(filename):
stop = 0.90
thres = 40
flag = True
read = MAR()
read = read.create(filename)
read.restart()
read = MAR()
read = read.create(filename)
target = int(read.get_allpos() * stop)
while True:
pos, neg, total = read.get_numbers()
# print("%d, %d" %(pos,pos+neg))
if pos >= target:
break
if pos == 0 or pos + neg < thres:
for id in read.random():
read.code(id, read.body["label"][id])
else:
a, b, ids, c = read.train(pne=True)
if pos >= 60 and flag:
read.cache_est()
# read.xx=read.simcurve['x']
# read.yy=read.simcurve['pos']
flag = False
for id in ids:
read.code(id, read.body["label"][id])
return read
def LINEAR(filename):
read = MAR()
read = read.create(filename)
while True:
pos, neg, total = read.get_numbers()
if total - (pos + neg) < 10:
break
for id in read.random():
read.code(id, read.body["label"][id])
return read
def LOC(filename):
stop = 1
read = MAR()
read = read.create(filename)
target = int(read.get_allpos() * stop)
while True:
pos, neg, total = read.get_numbers()
print("%d, %d" % (pos, pos + neg))
if pos >= target:
break
for id in read.loc_sort():
read.code(id, read.body["label"][id])
return read
def START_ERROR(filename):
read = MAR()
read = read.create(filename)
pos_last = 0
full_life = 3
human_error = 0.2
life = full_life
while True:
pos, neg, total = read.get_numbers()
print("%d/ %d" % (pos, pos + neg))
if pos >= 10:
if pos == pos_last:
life = life - 1
if life == 0:
break
else:
life = full_life
if pos == 0:
for id in read.random():
if read.body["label"][id] == "no":
if random.random() < human_error**2:
hl = "yes"
else:
hl = "no"
elif read.body["label"][id] == "yes":
if random.random() < 2 * (human_error - human_error**2):
hl = "no"
else:
hl = "yes"
read.code(id, hl)
else:
a, b, ids, c = read.train()
for id in ids:
if read.body["label"][id] == "no":
if random.random() < human_error**2:
hl = "yes"
else:
hl = "no"
elif read.body["label"][id] == "yes":
if random.random() < 2 * (human_error - human_error**2):
hl = "no"
else:
hl = "yes"
read.code(id, hl)
pos_last = pos
read.export()
return read
def REUSE_RANDOM(filename, old):
stop = 0.9
read = MAR()
read = read.create(filename)
read.create_old(old)
num2 = read.get_allpos()
target = int(num2 * stop)
while True:
pos, neg, total = read.get_numbers()
# print("%d/ %d" % (pos,pos+neg))
if pos >= target:
break
a, b, ids, c = read.train_reuse_random()
for id in ids:
read.code(id, read.body["label"][id])
return read
def UPDATE(filename, old, pne=False, cl="RF"):
stop = 1
read = MAR()
read = read.create(filename)
read.create_old(old)
num2 = read.get_allpos()
target = int(num2 * stop)
while True:
pos, neg, total = read.get_numbers()
print("%d/ %d" % (pos, pos + neg))
if pos >= target:
break
a, b, ids, c = read.train(pne=pne, cl=cl)
for id in ids:
read.code(id, read.body["label"][id])
return read
def START_LOC(filename, cl="SVM-linear"):
stop = 1
read = MAR()
read = read.create(filename)
target = int(read.get_allpos() * stop)
while True:
pos, neg, total = read.get_numbers()
print("%d, %d" % (pos, pos + neg))
if pos >= target:
break
if pos == 0 or pos + neg < 40:
for id in read.loc_sort():
read.code(id, read.body["label"][id])
else:
a, b, ids, c = read.train(cl=cl)
for id in ids:
read.code(id, read.body["label"][id])
return read
def TIME_START(filename):
stop = 0.9
read = MAR()
read = read.create(filename)
num2 = read.get_allpos()
target = int(num2 * stop)
while True:
pos, neg, total = read.get_numbers()
# print("%d/ %d" % (pos,pos+neg))
if pos >= target:
break
if pos == 0:
for id in read.random():
read.code(id, read.body["label"][id])
else:
a, b, ids, c = read.train_kept()
for id in ids:
read.code(id, read.body["label"][id])
return read
def UPDATE_AUTO(filename, old, pne=True):
read = MAR()
read = read.create(filename)
read.create_old(old)
pos_last = -1
full_life = 5
life = full_life
while True:
pos, neg, total = read.get_numbers()
print("%d/ %d" % (pos, pos + neg))
if pos == pos_last:
life = life - 1
if life == 0:
break
else:
life = full_life
a, b, ids, c = read.train(pne)
for id in ids:
read.code(id, read.body["label"][id])
pos_last = pos
return read
def START_DOC2VEC(filename):
stop = 0.95
thres = 40
read = MAR()
read = read.create(filename)
read.restart()
read = MAR()
read = read.create(filename)
target = int(read.get_allpos() * stop)
while True:
pos, neg, total = read.get_numbers()
print("%d, %d" % (pos, pos + neg))
if pos >= target:
break
if pos == 0 or pos + neg < thres:
for id in read.random():
read.code(id, read.body["label"][id])
else:
a, b, c, d, e = read.train(weighting=True)
for id in c:
read.code(id, read.body["label"][id])
return read
def REUSE(filename, old, pne=True):
stop = 0.9
thres = 5
read = MAR()
read = read.create(filename)
read.create_old(old)
num2 = read.get_allpos()
target = int(num2 * stop)
while True:
pos, neg, total = read.get_numbers()
print("%d/ %d" % (pos, pos + neg))
if pos >= target:
break
if pos < thres:
a, b, ids, c = read.train(pne)
for id in ids:
read.code(id, read.body["label"][id])
else:
a, b, ids, c = read.train_reuse(pne)
for id in ids:
read.code(id, read.body["label"][id])
return read
def UPDATE_REUSE(filename, old):
stop = 0.9
lifes = 2
life = lifes
last_pos = 0
thres = 5
read = MAR()
read = read.create(filename)
read.create_old(old)
num2 = read.get_allpos()
target = int(num2 * stop)
while True:
pos, neg, total = read.get_numbers()
# print("%d/ %d" % (pos, pos + neg))
if pos - last_pos:
life = lifes
else:
life = life - 1
last_pos = pos
if pos >= target:
break
# if (pos >= thres or pos==0) and life<1:
if (pos >= thres) and life < 1:
# print("reuse")
lifes = 0
a, b, ids, c = read.train_reuse()
for id in ids:
read.code(id, read.body["label"][id])
else:
# print("update")
a, b, ids, c = read.train()
for id in ids:
read.code(id, read.body["label"][id])
return read
def Codes(filename, code):
stop = 0.95
thres = 0
if "P" in code:
starting = 5
else:
starting = 1
weighting = "W" in code or "M" in code
uncertain = "U" in code
stopping = "S" in code
read = MAR()
read = read.create(filename)
read.restart()
read = MAR()
read = read.create(filename)
if not ("A" in code or "M" in code):
read.enough = 100000
target = int(read.get_allpos() * stop)
while True:
pos, neg, total = read.get_numbers()
# print("%d, %d" %(pos,pos+neg))
if pos >= target:
break
if pos < starting or pos + neg < thres:
for id in read.random():
read.code(id, read.body["label"][id])
else:
a, b, c, d, e = read.train(weighting=weighting)
if pos < 30 and uncertain:
for id in a:
read.code(id, read.body["label"][id])
else:
if stopping:
now = 0
while pos < target:
for id in e[now:now + read.step]:
read.code(id, read.body["label"][id])
pos, neg, total = read.get_numbers()
now = now + read.step
else:
for id in c:
read.code(id, read.body["label"][id])
return read
read.BM25(query.split())
while True:
pos, neg, total = read.get_numbers()
try:
print("%d, %d, %d" % (pos, pos + neg, read.est_num))
except:
print("%d, %d" % (pos, pos + neg))
if pos + neg >= total:
break
if pos < 1:
if query:
ids, scores = read.BM25_get()
for id in ids:
read.code(id, read.body["label"][id])
else:
for id in read.random():
read.code(id, read.body["label"][id])
else:
uncertain, uncertain_proba, certain, certain_proba, _ = read.train(
weighting=True, pne=True)
if target_recall * read.est_num <= pos:
break
if pos <= thres:
for id in uncertain:
read.code(id, read.body["label"][id])
else:
for id in certain:
read.code(id, read.body["label"][id])