-
Notifications
You must be signed in to change notification settings - Fork 24
/
agency2vec.py
65 lines (55 loc) · 1.77 KB
/
agency2vec.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
import argparse
import json
import numpy as np
import matplotlib.pyplot as plt
from seaborn import scatterplot
from gensim.models import Word2Vec
def main(
channels_path,
clusters_path,
output_png
):
with open(channels_path) as r:
channels = json.load(r)
channels = {channel["name"]: channel for channel in channels}
with open(clusters_path) as r:
clusters = [json.loads(line) for line in r]
texts = []
for cluster in clusters:
cluster_channels = {doc["channel_id"].lower() for doc in cluster["docs"]}
texts.append(list(cluster_channels))
model = Word2Vec(
sentences=texts,
vector_size=2,
window=15,
min_count=1,
workers=16,
sg=0,
epochs=60,
hs=1,
negative=0
)
vectors = model.wv
keys = list(vectors.key_to_index.keys())
print(keys)
names = [channels[key]["alias"] for key in keys]
colors = [channels[key]["group"] for key in keys]
matrix = np.zeros((len(keys), 2), dtype=np.float)
for i, key in enumerate(keys):
for j, elem in enumerate(vectors[key]):
matrix[i, j] = elem
x = matrix[:, 0]
y = matrix[:, 1]
plt.figure(figsize=(8, 8), dpi=100)
scatterplot(x, y, c=colors)
plt.title("Agency2Vec на постах канала")
for point_x, point_y, name in zip(x, y, names):
plt.text(point_x + 0.015, point_y + 0.015, name)
plt.savefig("agency2vec.png")
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--channels-path', type=str, required=True)
parser.add_argument('--clusters-path', type=str, required=True)
parser.add_argument('--output-png', type=str, required=True)
args = parser.parse_args()
main(**vars(args))