def __init__(self,
config,
model_path='./runs/1539678339/checkpoints/model-200',
word_to_index='./vocabs/word_to_index.json',
index_to_label='./vocabs/index_to_label.json'):
self.word_to_index = load_json(word_to_index)
self.index_to_label = load_json(index_to_label)
graph = tf.Graph()
with graph.as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=config['allow_soft_placement'],
log_device_placement=config['log_device_placement'])
self.sess = tf.Session(config=session_conf)
with self.sess.as_default():
# Load the saved meta graph and restore variables
saver = tf.train.import_meta_graph(
"{}.meta".format(model_path))
saver.restore(self.sess, model_path)
# Get the placeholders from the graph by name
self.input_x = graph.get_operation_by_name(
"input_x").outputs[0]
self.dropout_keep_prob = graph.get_operation_by_name(
"dropout_keep_prob").outputs[0]
# Tensors we want to evaluate
self.predictions = graph.get_operation_by_name(
"output/predictions").outputs[0]
def __init__(self,
config,
model_path='./runs/1548754630/checkpoints/model-1500',
word2index='./vocabs/word2index.json',
index2label='./vocabs/index2label.json'):
self.word2index = load_json(word2index)
self.index2label = load_json(index2label)
graph = tf.Graph()
with graph.as_default():
session_conf = tf.ConfigProto(
allow_soft_placement=config['allow_soft_placement'],
log_device_placement=config['log_device_placement'])
self.sess = tf.Session(config=session_conf)
with self.sess.as_default(): # 使用as_default(),当退出上下文的时候会话不会关闭
# load model
saver = tf.train.import_meta_graph(
'{}.meta'.format(model_path))
saver.restore(self.sess, model_path)
# get the placeholders from graph by name
self.input_x = graph.get_operation_by_name(
'input_x').outputs[0]
self.dropout_keep_prob = graph.get_operation_by_name(
'dropout_keep_prob').outputs[0]
# tensors we want to evaluate
self.predictions = graph.get_operation_by_name(
'output/predictions').outputs[0]
def generating_training_time_series(lower_range, upper_range):
# Read stream data
_main_stream = dh.load_json(PATH + 'james.json') # All main stream data.
#reply_stream = dh.load_json('james_reply.json') # All reply stream data.
_reply_stream = [
dh.load_json(PATH + "james/{}.json".format(i))
for i in range(len(_main_stream))
]
# main posts time series
main_stream_time_series = dh.main_stream_time_lists(
_main_stream)[lower_range:upper_range]
reply_stream_time_series = dh.reply_stream_time_lists(
_reply_stream)[lower_range:upper_range]
return main_stream_time_series, reply_stream_time_series
def load_data_all():
# glyce embedding
fin = "../data/wvec.hf"
init_vec = load_hdf(fin)
# char vec from training
fin = "../data/trn_wvec.hf"
train_vec = load_hdf(fin)
# glyce char2vec idx
fdic = "../data/dictionary.json"
fdic = load_json(fdic)
init_ch2idx = fdic["char2idx"]
# char2vec idx from training
ch2idx = load_json("../md/char2idx.json")
return init_ch2idx, init_vec, ch2idx, train_vec
def doc2vec(fci):
dic = load_json(fci)
init_ch2idx, init_vec, ch2idx, train_vec = load_data_all()
raw_dic, dic = make_data(dic, ch2idx, train_vec, init_ch2idx, init_vec)
fout = "../data/char2vec_with_glyce.pickle"
write2pickle(fout, dic)
fout = "../data/char2vec_raw.pickle"
write2pickle(fout, raw_dic)
def generating_reply_time_series(upper_range, number):
# Read stream data
_main_stream = dh.load_json(PATH + 'james.json') # All main stream data.
#reply_stream = dh.load_json('james_reply.json') # All reply stream data.
_reply_stream = [
dh.load_json(PATH + "james/{}.json".format(i))
for i in range(len(_main_stream))
]
# main posts time series
main_stream_time_series = dh.main_stream_time_lists(
_main_stream)[upper_range:upper_range + number]
reply_stream_time_series = dh.reply_stream_time_lists(
_reply_stream)[upper_range:upper_range + number]
for i in range(len(reply_stream_time_series)):
temp = np.array(reply_stream_time_series[i])
reply_stream_time_series[i] = temp[np.where(
temp < main_stream_time_series[-1])]
return main_stream_time_series, reply_stream_time_series
text_list[i] = ''
return text_list
def searchkw(text_list, kw_lists):
"""Search for targeted elements in the list based on the lw_lists"""
mentions_list = []
for i, x in enumerate(text_list):
if any(n in x.lower() for n in kw_lists):
mentions_list.append(i)
return mentions_list
main_stream = dh.load_json("data/Avengers/avengers.json")
texts = generate_text(main_stream)
mentions = searchkw(texts, ['lbj', 'lebron', 'lebron james'])
# Get all posts related to James
james_dataframe = main_stream.iloc[mentions]
james_main_id = list(james_dataframe.sub_id)
# Generate James related
james_reply_stream = pd.DataFrame([])
for i in range(8):
df = dh.load_json('data/NBA_1904/reply_stream_{}.json'.format(i+1))
for i in range(len(df)):
if df.iloc[i].link_id in james_main_id:
