def topics(k):
from topicexplorer.lib.color import rgb2hex
import numpy as np
response.content_type = 'application/json; charset=UTF8'
response.set_header('Expires', _cache_date())
response.set_header('Cache-Control', 'max-age=86400')
# set a parameter for number of words to return
wordmax = 10 # for alphabetic languages
if kwargs.get('lang', None) == 'cn':
wordmax = 25 # for ideographic languages
# populate word values
phi = self.v[k].phi.T
idxs = phi.argsort(axis=1)[:, ::-1][:, :wordmax]
# https://github.com/numpy/numpy/issues/4724
idx_hack = np.arange(np.shape(phi)[0])[:, np.newaxis]
dt = [('Word', self.c.words.dtype), ('Prob', phi.dtype)]
data = np.zeros(shape=(phi.shape[0], wordmax), dtype=dt)
data['Word'] = self.c.words[idxs]
data['Prob'] = phi[idx_hack, idxs]
js = {}
for i, topic in enumerate(data):
js[str(i)] = {
"color": rgb2hex(self.colors[k][i]),
'words':
dict([(str(w), float(p)) for w, p in topic[:wordmax]])
}
return json.dumps(js)
def topics(k):
from topicexplorer.lib.color import rgb2hex
import numpy as np
etag = _generate_etag(self.v[k])
# Check if there is a "If-None-Match" ETag in the request
if request.get_header('If-None-Match', '') == etag:
response.status = 304
return "Not Modified"
if k not in self.topic_range:
response.status = 400 # Not Found
return "No model for k = {}".format(k)
response.content_type = 'application/json; charset=UTF8'
response.set_header('Expires', _cache_date())
response.set_header('Cache-Control', 'max-age=120')
response.set_header('ETag', etag)
# set a parameter for number of words to return
wordmax = 10 # for alphabetic languages
if kwargs.get('lang', None) == 'cn':
wordmax = 25 # for ideographic languages
# populate word values
phi = self.v[k].phi.T
idxs = phi.argsort(axis=1)[:, ::-1][:, :wordmax]
# https://github.com/numpy/numpy/issues/4724
idx_hack = np.arange(np.shape(phi)[0])[:, np.newaxis]
dt = [('Word', self.c.words.dtype), ('Prob', phi.dtype)]
data = np.zeros(shape=(phi.shape[0], wordmax), dtype=dt)
data['Word'] = self.c.words[idxs]
data['Prob'] = phi[idx_hack, idxs]
labels = []
if self.label_file:
with open(self.label_file) as labels_in:
for label in labels_in:
label = label.strip()
labels.append(label)
else:
for i, _ in enumerate(data):
labels.append('Topic {}'.format(i))
js = {}
for i, topic in enumerate(data):
js[text(i)] = {
"color":
rgb2hex(self.colors[k][i]),
'words':
dict([(text(w), float(p)) for w, p in topic[:wordmax]]),
'label':
labels[i]
}
return json.dumps(js)
def topics(k):
from topicexplorer.lib.color import rgb2hex
import numpy as np
etag = _generate_etag(self.v[k])
# Check if there is a "If-None-Match" ETag in the request
if request.get_header('If-None-Match', '') == etag:
response.status = 304
return "Not Modified"
if k not in self.topic_range:
response.status = 400 # Not Found
return "No model for k = {}".format(k)
response.content_type = 'application/json; charset=UTF8'
response.set_header('Expires', _cache_date())
response.set_header('Cache-Control', 'max-age=120')
response.set_header('ETag', etag)
# set a parameter for number of words to return
wordmax = 10 # for alphabetic languages
if kwargs.get('lang', None) == 'cn':
wordmax = 25 # for ideographic languages
# populate word values
phi = self.v[k].phi.T
idxs = phi.argsort(axis=1)[:,::-1][:,:wordmax]
# https://github.com/numpy/numpy/issues/4724
idx_hack = np.arange(np.shape(phi)[0])[:,np.newaxis]
dt = [('Word',self.c.words.dtype),('Prob',phi.dtype)]
data = np.zeros(shape=(phi.shape[0], wordmax), dtype=dt)
data['Word'] = self.c.words[idxs]
data['Prob'] = phi[idx_hack, idxs]
js = {}
for i, topic in enumerate(data):
js[text(i)] = {
"color": rgb2hex(self.colors[k][i]),
'words': dict([(text(w), float(p))
for w, p in topic[:wordmax]])
}
return json.dumps(js)
def topics(k):
from topicexplorer.lib.color import rgb2hex
response.content_type = 'application/json; charset=UTF8'
response.set_header('Expires', _cache_date())
response.set_header('Cache-Control', 'max-age=86400')
# populate word values
data = self.v[k].topics()
js = {}
wordmax = 10 # for alphabetic languages
if kwargs.get('lang', None) == 'cn':
wordmax = 25 # for ideographic languages
for i, topic in enumerate(data):
js[str(i)] = {
"color": rgb2hex(self.colors[k][i]),
'words': dict([(str(w), float(p))
for w, p in topic[:wordmax]])
}
return json.dumps(js)
def topics():
from topicexplorer.lib.color import rgb2hex
response.content_type = 'application/json; charset=UTF8'
response.set_header('Expires', _cache_date())
# populate partial jsd values
data = lda_v.topic_jsds()
js = {}
for rank,topic_H in enumerate(data):
topic, H = topic_H
js[str(topic)] = {
"H" : H,
"color" : rgb2hex(colors[topic])
}
# populate word values
data = lda_v.topics()
for i,topic in enumerate(data):
js[str(i)].update({'words' : dict([(w, p) for w,p in topic[:10]])})
return json.dumps(js)
def topics():
from topicexplorer.lib.color import rgb2hex
response.content_type = 'application/json; charset=UTF8'
response.set_header('Expires', _cache_date())
# populate partial jsd values
data = lda_v.topic_jsds()
js = {}
for rank,topic_H in enumerate(data):
topic, H = topic_H
js[str(topic)] = {
"H" : H,
"color" : rgb2hex(colors[topic])
}
# populate word values
data = lda_v.topics()
for i,topic in enumerate(data):
js[str(i)].update({'words' : dict([(w, p) for w,p in topic[:10]])})
return json.dumps(js)