def prepare_set_of_lexicon():
# Step 1: generate lexicon
lexicon = lex.generate_lexicon_from_textfile(TRAINING_FILE)
lex.write_lexicon_to_file(LEX_FILE, lexicon)
cutoff_lexicon = lex.generate_cutoff_lexicon(lexicon, './files_from_outside/english.stop.txt')
lex.write_lexicon_to_file(CUTOFF_LEX_FILE, cutoff_lexicon)
refined_cutoff_lexicon = lex.generate_refined_cutoff_lexicon(lexicon, './files_from_outside/english.stop.txt')
lex.write_lexicon_to_file(REFINED_CUTOFF_LEX_FILE, refined_cutoff_lexicon)
for (class_name, concepts) in self.class_and_concept.items():
for concept in concepts:
output.write('%d\t%d\t%s\t%s\n' % (0, 0, class_name, concept))
# <unk> to every possible concepts
for concept in self.concepts:
output.write('%d\t%d\t%s\t%s\n' % (0, 0, '<unk>', concept))
# word_without_associated_class to the same word
for word in self.word_without_associated_class:
output.write('%d\t%d\t%s\t%s\n' % (0, 0, word, word))
# write last line
output.write('0')
if __name__ == '__main__':
w2concept = W2Concept(CONCEPT_FILENAME)
w2concept.set_word_without_associated_class('./w2class/word_without_class.txt')
w2concept.write_w2concept_transducer('./w2concept/w2concept.fsm')
lexicon = lex.read_lexicon_file(LEX_FILE)
new_lexicon = w2concept.concepts.union(set(w2concept.class_and_concept.keys()))
new_lexicon = new_lexicon.union(w2concept.word_without_associated_class)
lexicon = lexicon.union(new_lexicon)
lex.write_lexicon_to_file(LEX_FILE, list(lexicon))
def update_lexicon(self, filename):
lexicon = self.lexicon.union(self.class_name)
lex.write_lexicon_to_file(filename, list(lexicon))
self.write_w2class_transducer(self.symbols, '')
def read_base_lex(filename):
lexicon = []
with open(filename) as f:
for line in f:
lexicon.append(line.strip())
return lexicon
def write_lexicon_to_file(filename, lexicon):
"""
set(): lexicon
"""
# remove empty string character
try:
lexicon.remove('')
except:
pass
count = 0
special_lex = ['<epsilon>', '<unk>']
with open(filename, 'w') as output:
for lex in special_lex + lexicon:
output.write('%s\t%d\n' % (lex, count))
count += 1
if __name__ == '__main__':
lexicon = lex.generate_lexicon_from_textfile(TRAINING_FILE)
lex.write_lexicon_to_file(LEX_FILE, lexicon)
