Example #1
0
):  #This loop from https://github.com/jaimezorno/Deep-Learning-for-NLP-Creating-a-Chatbot/blob/master/Deep%20Learning%20for%20NLP-%20Creating%20a%20chatbot.ipynb
    if val == val_max:
        k = key
print(k)
print(pred_res[0][val_max])


#Here is where a story and question is input line by line
def input_story():
    story = []
    for i in range(3):
        story.append(
            input("Enter Story Sentence :")
        )  # note: Add a space after the final period, ie. 'apple . '
    question = input("What is the question?:")
    return ''.join(word for word in story), question


story, question = input_story()
my_s = [(story.split(), question.split(), 'yes')]
print(my_s)
the_story, the_question, the_ans = vectorize_stories(my_s)
pred_res = model.predict(([the_story, the_question]))
val_max = np.argmax(pred_res[0])
for key, val in t.word_index.items(
):  #This loop from https://github.com/jaimezorno/Deep-Learning-for-NLP-Creating-a-Chatbot/blob/master/Deep%20Learning%20for%20NLP-%20Creating%20a%20chatbot.ipynb
    if val == val_max:
        k = key
print(k)
print(pred_res[0][val_max])
Example #2
0
import numpy as np
import pandas as pd
from keras.preprocessing.text import Tokenizer
from collections import Counter
from keras.layers import Embedding, Input, LSTM, Dense, Conv1D, Conv2D, MaxPool2D, MaxPooling1D, Dropout, Activation, Reshape, Concatenate, Flatten
from keras.models import Sequential, model_from_json
from keras.models import Model
from keras.preprocessing.sequence import pad_sequences
from keras.utils import np_utils

# read the input
f = open("LabelledData.txt", 'r')
Input = f.read()
Input = Input.split("\n")
Input_len = len(Input)
Input_data = {}
# seperating label and data
InputData = {}
InputLabel = {}
for j in range(0, Input_len - 2):
    b = Input[j].split(",,, ")
    Input_data[j] = b
    InputData[j] = Input_data[j][0]
    InputLabel[j] = Input_data[j][1]

d = InputLabel.values()
tokenizerlabel = Tokenizer(num_words=20000)
tokenizerlabel.fit_on_texts(d)
sequenceslabel = tokenizerlabel.texts_to_sequences(d)

# As the data has imbalanced classifiers