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
66
67
68
69
70
71
72

#! /usr/bin/env python

import tensorflow as tf
import numpy as np
import os
import time
import datetime
import data_helpers
from text_cnn import TextCNN

from flask import Flask, request, render_template

print("Loading data...")
_, _, vocabulary, vocabulary_inv = data_helpers.load_data()


tcnn = TextCNN(sequence_length=119, num_classes=2, vocab_size=40000, embedding_size=128, filter_sizes=[3,4,5], num_filters=128)

sess = tf.Session()
def predict(_input):
 ret = "negative"
 if 1 == sess.run(tcnn.predictions, {tcnn.input_x: _input, tcnn.dropout_keep_prob: 1.0})[0] :
  ret = "positive"
 return ret 
 
saver = tf.train.Saver()
saver.restore(sess, "/root/cnn_sentiment_classification/runs/1477243665/checkpoints/model-46800")

app = Flask(__name__)

@app.route('/', methods=['GET', 'POST'])
def index():
 if request.method == 'POST':
  s = request.form["sentence"]
  original_sentence = s
  s = data_helpers.clean_str(s)
  s = s.split(" ")
  # 119 is max str len
  s = s + ["<PAD/>"] * (119 - len(s))
  for i, w in enumerate(s):
   if w not in vocabulary:
    s[i] =  "<PAD/>"
  s = np.array([vocabulary[word] for word in s])
  s = [s]
  result = predict(s)
  return render_template("index.html", result=result, original_sentence=original_sentence) 
 
 return render_template("index.html")

if __name__ == '__main__':
 app.run(host='0.0.0.0')


# ##################################################################################################
"""
while True:
 s = raw_input("type the sentence:")
 if s == "end":
  break;
 s = data_helpers.clean_str(s)
 s = s.split(" ")
 # 119 is max str len
 s = s + ["<PAD/>"] * (119 - len(s))
 for i, w in enumerate(s):
  if w not in vocabulary:
   s[i] =  "<PAD/>"
 s = np.array([vocabulary[word] for word in s])
 s = [s]

 print "Prediction: ", sess.run(tcnn.predictions, {tcnn.input_x: s, tcnn.dropout_keep_prob: 1.0})
"""
# ###########