-
Notifications
You must be signed in to change notification settings - Fork 0
/
model.py
207 lines (135 loc) · 5.26 KB
/
model.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
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
# -*- coding: utf-8 -*-
"""
Created on Tue Sep 28 18:02:34 2021
@author: Dhakshin Krishna J & Team
"""
#There is a massive earthquake in Nepal, which has caused a lot of damage to the surroundings around the epicenter
#To train and save the model
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split
from keras.preprocessing.sequence import pad_sequences
from tensorflow.keras.utils import to_categorical
from tensorflow.keras import Model,Input
from tensorflow.keras.layers import LSTM,Embedding,Dense
from tensorflow.keras.layers import TimeDistributed, SpatialDropout1D,Bidirectional
plt.style.use('seaborn')
#setting function
def setting(mode):
if mode=='0':
#Advanced mode
modifier=1
tagTable=1
tokenArray=1
elif mode == '1':
#Viewer mode
modifier=0
tagTable=0
tokenArray=1
elif mode=='2':
#Modifier mode
modifier=1
tagTable=0
tokenArray=1
else:
#Default mode
modifier=0
tagTable=1
tokenArray=0
return modifier,tagTable,tokenArray
#combining two csv files to create the dataframe
TheDataList=[]
listOfCsv=["datasets/ner_datasetreference.csv","datasets/added.csv"]
for x in listOfCsv:
TheDataList.append(pd.read_csv(x,encoding='unicode_escape',skipinitialspace=True,skip_blank_lines=True))
data=pd.concat(TheDataList)
data.head(-20)
data=data.fillna(method="ffill")
data.head(-1)
words=list(set(data['Word'].values))
words.append("ENDPAD")
#print(words)
num_words=len(words)
print("Total number of words",num_words)
tags = list(set(data["Tag"].values))
tags.sort()
num_tags = len(tags)
print("List of tags: " + ', '.join([tag for tag in tags]))
print(f"Total Number of tags {num_tags}")
#creating class for the model to access
class Get_sentence(object):
def __init__(self,data):
self.n_sent=1
self.data=data
agg_func=lambda s:[(w,t,p) for w,t,p in zip(s['Word'].tolist(),s['Tag'].tolist(),s['POS'].tolist())]
self.grouped=self.data.groupby('Sentence #').apply(agg_func)
self.sentences=[s for s in self.grouped]
#creating object of the model class...
getter=Get_sentence(data)
sentence=getter.sentences
for s in sentence:
len_s=len(s)
#initialize id for words and tags
word_idx = {w : i + 1 for i ,w in enumerate(words)}
tag_idx = {t : i for i ,t in enumerate(tags)}
print(tag_idx)
# word_idx["India"]
# tag_idx["Dis"]
# tag_idx
#plot figure for the frequency of length of sentences
plt.figure(figsize=(14,7))
plt.hist([len(s) for s in sentence],bins = 50)
plt.xlabel("Length of Sentences")
plt.show()
#plot the figure for frequency of tags
plt.figure(figsize=(14, 7))
plt.xlabel("Frequency of tags")
data.Tag[data.Tag != 'O'].value_counts().plot.barh();
# word_idx.keys()
#checking max-length of sentences and padding every sentence to max length
max_len=max([len(s) for s in sentence])
# max_len
X=[[word_idx[w[0]] for w in s] for s in sentence]
X=pad_sequences(maxlen=max_len,sequences=X,padding='post',value=num_words)
y=[[tag_idx[w[1]]for w in s]for s in sentence]
y=pad_sequences(maxlen=max_len,sequences=y,padding='post',value=tag_idx['O'])
y=[to_categorical(i,num_classes=num_tags) for i in y]
X_train,X_test,y_train,y_test=train_test_split(X,y,test_size=0.1,random_state=1)
#np.shape(X_train)
#np.shape(X_test)
#np.shape(y_test)
#np.shape(y_train)
input_word=Input(shape=(max_len,))
# model layering using BI-lstm
model=Embedding(input_dim=num_words+1,output_dim=max_len,input_length=max_len)(input_word)
model=SpatialDropout1D(0.1)(model)
model=Bidirectional(LSTM(units=52,return_sequences=True,recurrent_dropout=0.1))(model)
out=TimeDistributed(Dense(num_tags,activation='softmax'))(model)
model=Model(input_word,out)
model.compile(optimizer='adam',loss='categorical_crossentropy',metrics=['accuracy'])
model.summary()
#fitting model and splitting the test/train data in 20:80 ration
model.fit(X_train,np.array(y_train),verbose=1,epochs=1,validation_split=0.2)
print("------------------------------Model Training complete------------")
model.evaluate(X_test,np.array(y_test))
print("-------------------------------Model Testing complete------------")
model.save("savedModel");
print("-------------------------------Model Stored for future------------")
#RANODMLY PICK A SENTENCE AND TEST THE OUTCOME
x=input("Want to test a random sentence from the dataset?y/n:")
if x=='y' or x=='Y':
rand_sent=np.random.randint(0,X_test.shape[0])
p=model.predict(np.array([X_test[rand_sent]]))
p=np.argmax(p,axis=-1)
y_true=np.argmax(np.array(y_test),axis=-1)[rand_sent]
print("{:20}{:20}\t{}\n".format("Word","Truth","Pred"))
print("-"*55)
strl=" "
for (w,t,pred)in zip(X_test[rand_sent],y_true,p[0]):
if(words[w-1] != "ENDPAD"):
if not(tags[t]=="O" and tags[pred]=="O"):
print("{:20}{:20}\t{}".format(words[w-1],tags[t],tags[pred]))
if(words[w-1] != "ENDPAD"):
strl=strl+words[w-1]+" "
print("\n\n"+strl);