def _makeTextMatrix(self, inputFile, stopwordFile):
# Generate text matrix with TmgSimple from 02450
textMatrix = TmgSimple(filename=formattedDatabase,
stopwords_filename=stopwordFile)
attributeNames = textMatrix.get_words(sort=True)
# Make an output file
attFile = open(attributFile, 'w')
datFile = open(dataFile, 'w')
for word in attributeNames:
attFile.write(word)
attFile.write('\n')
attFile.close
for i in range(40):
np.savetxt(datFile, textMatrix.get_matrix(i*1000, (i+1)*1000,
sort=True), fmt='%i')
datFile.close
# exercise 3.1.4
import numpy as np
from tmgsimple import TmgSimple
# Generate text matrix with help of simple class TmgSimple
tm = TmgSimple(filename='../Data/textDocs.txt',
stopwords_filename='../Data/stopWords.txt',
stem=True)
# Extract variables representing data
X = tm.get_matrix(sort=True)
attributeNames = tm.get_words(sort=True)
# Display the result
print(attributeNames)
print(X)
print('Ran Exercise 3.1.4')
bagOfWords = ['matrix', 'Google', 'ranking', 'web', 'webpage', 'rank']
"""
3.1.2
"""
import numpy as np
from tmgsimple import TmgSimple
from similarity import similarity
# Generate text matrix with help of simple class TmgSimple
tm = TmgSimple(filename='../02450Toolbox_Python/Data/textDocs.txt', )
# Extract variables representing data
X = tm.get_matrix(sort=True)
attributeNamesWithOutStop = tm.get_words(sort=True)
# Display the result
print attributeNamesWithOutStop
print X
"""
3.1.3
With stopwords
"""
print('Now with stopwords !!!')
tm = TmgSimple(filename='../02450Toolbox_Python/Data/textDocs.txt', stopwords_filename='../02450Toolbox_Python/Data/stopWords.txt')
# Extract variables representing data
X = tm.get_matrix(sort=True)
attributeNamesWithStop = tm.get_words(sort=True)
# exercise 2.1.4 import numpy as np from tmgsimple import TmgSimple # Generate text matrix with help of simple class TmgSimple tm = TmgSimple(filename='../Data/textDocs.txt', stopwords_filename='../Data/stopWords.txt', stem=True) # Extract variables representing data X = tm.get_matrix(sort=True) attributeNames = tm.get_words(sort=True) # Display the result print attributeNames print X