Sentence-Classification/w2v_utils.py at master · Nomiluks/Sentence-Classification · GitHub

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
from sklearn.metrics import precision_recall_fscore_support as score
from sklearn.metrics import classification_report
import sklearn
from nltk.corpus import stopwords
from tools import print_confusion_matrix
from bs4 import BeautifulSoup
import numpy as np
import pandas as pd
import os
import re
import random

def review_to_wordlist( review, remove_stopwords=True ):
    # Function to convert a document to a sequence of words,
    # optionally removing stop words.  Returns a list of words.
    #

    # 1. Remove HTML
    review_text = BeautifulSoup(review, "lxml").get_text()
    #review_text = review
    #
    # 2. Remove non-letters
    review_text = re.sub("[^a-zA-Z]"," ", review_text)

    #
    # 3. Convert words to lower case and split them
    words = review_text.lower().split()
    #
    # 4. Optionally remove stop words (false by default)
    if remove_stopwords:
        stops = set(stopwords.words("english"))
        words = [w for w in words if not w in stops]
    #
    # 5. Return a list of words
    return(words)

###**********combine datasets for word2vec training************###
def word2vecInput(train,test,gold,yelp, shuffle=True, remove_stopwords=True):
    #initialized train and gold test sentences
    goldSet  = gold["review"].tolist()
    trainSet = train["review"].tolist()
    testSet  = test["review"].tolist()

    trainSentences = yelp+trainSet+testSet
    print "Total Train Sentences: ",len(trainSentences)
    print "Total Gold Test Sentences: ", len(goldSet)

    #cleaning sentences
    trainingSentences = [review_to_wordlist(review,remove_stopwords) for review in trainSentences]
    goldSentences     = [review_to_wordlist(review,remove_stopwords) for review in goldSet]
    #vocablary = trainingSentences+goldSentences
    #print "Total Vocablary of Sentences: ", len(vocablary)

    #Shuffle training sentences Sentences
    if (shuffle):
        random.shuffle(trainingSentences)

    return trainingSentences#, vocablary

def splitDataset(examples, labels, split=0.8):
    data = zip(examples,labels)
    random.shuffle(data)
    data_length = int(round(split*len(data)))
    train = data[:data_length] #training set
    validation = data[data_length:]#validation set
    train = zip(*train)
    validation = zip(*validation)
    x_train = pd.Series(np.array(train[0]))
    y_train = pd.Series(np.array(train[1]))
    x_val = pd.Series(np.array(validation[0]))
    y_val = pd.Series(np.array(validation[1]))
    return x_train, y_train, x_val, y_val

def tokenizerCleaner(yelp, remove_stopwords=True):
    print "Total Train Sentences: ",len(yelp)
    yelp = [review_to_wordlist(review,remove_stopwords) for review in yelp]
    return yelp

def oneHotVectors(labels):
    labels = pd.Series(labels)
    labels = pd.get_dummies(labels)
    labels.to_csv("Labels.csv")
    labels = labels.values
    return labels

def categoriesToLabels(labels):
    labels = pd.Series(labels)
    labels = pd.get_dummies(labels)
    labels = labels.values #copying the vectors for each category [1...13] length
    item_index = np.where(labels[:]==1)
    labels = item_index[1]
    return labels

def nextBatch(X_train, y_train, batch_size):
    sample = random.sample(zip(X_train,y_train), batch_size)
    sample = zip(*sample)
    return sample[0], sample[1]

def weight_variable(fan_in, fan_out, filename, boolean=False):
    initial=0
    if (boolean):
        stddev = np.sqrt(2.0/fan_in)
        initial  = tf.random_normal([fan_in,fan_out], stddev=stddev)
    else:
        initial  = np.loadtxt(filename).astype(np.float32)
        #print initial.shape
    return tf.Variable(initial)

def resetModel():
    files = glob.glob('params/*')
    for f in files:
        os.remove(f)

def bias_variable(shape, filename, boolean=False):
    initial=0
    if (boolean):
        initial = tf.constant(0.1, shape=shape)
    else:
        initial  = np.loadtxt(filename).astype(np.float32)
        #print initial.shape
    return tf.Variable(initial)

def confusionMatrix(y_pred, y_actu):
    return print_confusion_matrix(y_pred,y_actu)

def normalize(probs):
    prob_factor = 1 / np.sum(probs)
    return [prob_factor * p for p in probs]

def clean(sentences, remove_stopwords=False):
    sentences = [review_to_wordlist(review,remove_stopwords) for review in sentences]
    return sentences