file = f.read()
test_sentences = file.split('\n')
# Build the parser with corpus_train
print('Building the parser...')
cyk_parser = CYK(corpus_train)
print('Done')
# Parsing of Evaluation sentences
print('Parsing...')
test_sentences_bis = []
with open('data/evaluation_data.parser.txt', 'w') as f:
for sentence in test_sentences:
parsed_sentence = cyk_parser.parse(sentence)
if parsed_sentence is not None:
test_sentences_bis.append(sentence)
f.write('%s\n' % parsed_sentence)
print('Done')
# Get accuracy
# Get sentences parsed by our parser
with open('data/evaluation_data.parser.txt', 'r') as f:
file = f.read()
parsed_sentences = file.split('\n')
# Remove first two and last brackets to use parser from PYEVALB
initial_parsed_sentences = []
parsed_sentences_final = []
# Building Parser
print("Building PCFG and Parser")
my_CYK_parser = CYK(corpus_train)
print("Done")
print("Start Parsing")
if args.test_sentence:
sent = args.test_sentence
print("#################")
print("Sentence: ")
print(sent + "\n")
print("Parsing")
my_parsing = my_CYK_parser.parse(sent)
if my_parsing is None:
print("Found no viable parsing.")
else:
print(my_parsing)
for sent in open(args.test_file):
print("#################")
print("Sentence: ")
print(sent + "\n")
print("Parsing")
my_parsing = my_CYK_parser.parse(sent)
if my_parsing is None:
print("Found no viable parsing.")
for idx_sentence in range(nb_test):
print("##############################")
real_parsing = real_parsings_test[idx_sentence]
sent = sentences_test[idx_sentence]
print("Sentence #" + str(idx_sentence))
print(sent + "\n")
print("Real Parsing")
print(real_parsing + "\n")
print("Our CYK Parsing")
tic = time.time()
my_parsing = my_CYK_parser.parse(sent, viz_oov=False)
if my_parsing is None:
print("Found no viable parsing.")
else:
print(my_parsing)
tac = time.time()
print("Done in " + str(round(tac - tic, 2)) + "sec\n")
with open('results/evaluation_data.parser_output', 'a') as f:
if my_parsing is None:
f.write("Found no viable parsing." + "\n")
else:
f.write(my_parsing + "\n")
if my_parsing is not None:
# EVALPB works if we remove first and last brackets of the SEQUOIA format and the extra spaces that come with it