pu = []
for name in u[user]:
print '@' + root + name
count = count + 1
if count < 6:
fin = open(root + name + '.txt', 'r')
lines = fin.readlines()
fin.close()
pu.append([line.strip('\n') for line in lines])
else:
if count == 6 or type(zipf) == float and zipf < 1:
if stype == 3 and type(zipf) == float and zipf < 0:
result = sm.dg3(root + name, cll, b, s, p, wtolu, ltow,
ukk)
else:
result = sm.dg(root + name, cll, sys.argv[3], b, s,
wtolu, ukk, zipf, stype)
else:
if stype == 3 and type(zipf) == float and zipf < 0:
result = sm.dg4(root + name, cll, b, s, p, wtolu, ltow,
ukk, vecu)
else:
result = sm.dg2(root + name, cll, sys.argv[3], b, s,
wtolu, ukk, zipf, vecu, stype)
dlu = np.array([10.0 for i in range(0, len(result) - 1)])
mt = []
sim = []
for i in range(0, len(result) - 1):
if count == 6:
vect[i] = sm.vecof(result[i], a, wtola, akk)
mt.append(vect[i].max())
sim.append(attack.simatt(result[i], pu))
if len(u[user]) > 5:
print user
usernum = usernum + 1
count = 0
vec = [np.zeros(akk) for i in range(0, 4)]
vect = [np.zeros(akk) for i in range(0, 4)]
for name in u[user]:
count = count + 1
fin = open(root + name, 'r')
line = fin.read()
title = re.search(r'【発明の名称】(.*?)\(', line, re.DOTALL)
cl = re.search(r'(【国際特許分類第.*版】.*?)([A-H][0-9]+?[A-Z])', line,
re.DOTALL)
#print title.group(1)
#print cl.group(2)
result = sm.dg(root + name, cll, sys.argv[3], a, s, wtol, ukk,
zipf, type)
dl = np.array([10.0 for i in range(0, len(result) - 1)])
print '@' + root + name
for i in range(0, len(result) - 1):
if count == 1:
vect[i] = sm.vecof(result[i], a, wtol, akk)
else:
for j in range(0, len(result) - 1):
vecr = sm.vecof(result[i], a, wtol, akk)
dlt = spatial.distance.cosine(vecr, vec[j])
#print i,j,vecr.argmax(),dlt
if (dlt < dl[i]):
dl[i] = dlt
vect[i] = vec[j]
vect[i] = (vecr + vect[i] * count) / (count + 1)