def main():
while 1 == 1:
print("Enter a statement")
statement = raw_input()
statement = statement.lower()
tagged_arr = Viterbi(statement)
#check if all of the elements are same
count = 1
tag = tagged_arr[1]
for i in range(2, tagged_arr.__len__()):
if tagged_arr[i] == tag:
count = count + 1
if count == tagged_arr.__len__() - 1:
tokens = word_tokenize(statement)
n = tokens.__len__()
for i in range(0, n):
tag_temp = Viterbi(tokens[i])[1]
tagged_arr[i + 1] = tag_temp
if tokens[i] == 'open':
tagged_arr[i + 1] = 'VB'
if tokens[i] == 'file':
tagged_arr[i + 1] = 'NN'
print(tagged_arr)
CKY(tagged_arr)
"Prep -> 'to' | 'with'",
"Vi -> 'ate'", #intransitive
"Vt -> 'ate' | 'book' | 'Book' | 'gave' | 'told'", #transitive
"Vdt -> 'gave' | 'told' ", #ditransitive
"Subconj -> 'that'", #subordinating conjunction
"Mod -> 'Can' | 'will'", #modal verbs
"Aux -> 'did' ", #auxiliary verbs
"WhAdv -> 'Why'",
"PropN -> 'John' | 'Mary' | 'NYC' | 'London'",
"Adj -> 'nice' | 'drawing'",
"Pro -> 'you' | 'he'",
"Adv -> 'today'"
])
print(grammar) #the simpler grammar
chart=CKY(grammar)
#this illustrates tracing of a very simple sentence; feel free to try others.
chart.recognise(tokenise("the frogs swim"),True)
chart.pprint()
#build a chart with the larger grammar
chart2=CKY(grammar2)
# Note, please do _not_ use the Tree.draw() method uncommented
# _anywhere in this file_ (you are encouraged to use it in preparing
# your report).
# The sentences to examine.
#
for s in ["John gave a book to Mary.",
"John gave Mary a book.",
NP -> Det Nom | Nom | NP PP
Det -> NP "'s"
Nom -> N SRel | N
VP -> Vi | Vt NP | VP PP
PP -> Prep NP
SRel -> Relpro VP
Det -> 'a' | 'the'
N -> 'fish' | 'frogs' | 'soup' | 'children' | 'books'
Prep -> 'in' | 'for'
Vt -> 'saw' | 'ate' | 'read'
Vi -> 'fish' | 'swim'
Relpro -> 'that'
""")
print(grammar)
chart = CKY(grammar)
chart.recognise("the frogs swim".split()) # Should use
# tokenise(s) once that's fixed
chart.pprint()
# Q1: Uncomment this once you've completed Q1
chart.recognise(tokenise("the frogs swim"), True)
# Q3 Uncomment the next three once when you're working on Q3
chart.recognise(tokenise("fish fish"))
chart.pprint()
chart.recognise(tokenise("fish fish"), True)
# Use this grammar for the rest of the assignment
grammar2 = parse_grammar([
"S -> Sdecl '.' | Simp '.' | Sq '?' ", "Sdecl -> NP VP", "Simp -> VP",
"Sq -> Sqyn | Swhadv", "Sqyn -> Mod Sdecl | Aux Sdecl",
def single_layer_parser(self, trainer, diora, info):
logger = self.logger
epoch = info.get('epoch', 0)
original_K = diora.K
if self.K is not None:
diora.safe_set_K(self.K)
# set choose_tree
if hasattr(diora, 'choose_tree'):
original_choose_tree = diora.choose_tree
diora.choose_tree = self.choose_tree
word2idx = self.dataset['word2idx']
if self.cky_mode == 'cky':
parse_predictor = CKY(net=diora,
word2idx=word2idx,
add_bos_token=trainer.net.add_bos_token,
add_eos_token=trainer.net.add_eos_token)
elif self.cky_mode == 'diora':
parse_predictor = TreesFromDiora(diora=diora,
word2idx=word2idx,
outside=self.outside,
oracle=self.oracle)
batches = self.batch_iterator.get_iterator(random_seed=self.seed,
epoch=epoch)
logger.info('Parsing.')
pred_lst = []
counter = 0
eval_cache = {}
if self.ground_truth is not None:
self.ground_truth = os.path.expanduser(self.ground_truth)
ground_truth_data = {}
with open(self.ground_truth) as f:
for line in f:
ex = json.loads(line)
ground_truth_data[ex['example_id']] = ex
# Eval loop.
with torch.no_grad():
for i, batch_map in enumerate(batches):
batch_size, length = batch_map['sentences'].shape
if length <= 2:
continue
example_ids = batch_map['example_ids']
if self.ground_truth is not None:
batch_ground_truth = [
ground_truth_data[x] for x in example_ids
]
batch_map['ground_truth'] = batch_ground_truth
_ = trainer.step(batch_map,
train=False,
compute_loss=False,
info={
'inside_pool': self.inside_pool,
'outside': self.outside
})
for j, x in enumerate(parse_predictor.predict(batch_map)):
pred_lst.append(x)
self.eval_loop_hook(trainer, diora, info, eval_cache,
batch_map)
self.post_eval_hook(trainer, diora, info, eval_cache)
diora.safe_set_K(original_K)
# set choose_tree
if hasattr(diora, 'choose_tree'):
diora.choose_tree = original_choose_tree
return pred_lst
NP -> Det Nom | Nom | NP PP
Det -> NP "'s"
Nom -> N SRel | N
VP -> Vi | Vt NP | VP PP
PP -> Prep NP
SRel -> Relpro VP
Det -> 'a' | 'the'
N -> 'fish' | 'frogs' | 'soup' | 'children' | 'books'
Prep -> 'in' | 'for'
Vt -> 'saw' | 'ate' | 'read'
Vi -> 'fish' | 'swim'
Relpro -> 'that'
""")
print grammar
chart=CKY(grammar)
chart.parse("the frogs swim".split()) # Should use
# tokenise(s) once that's fixed
chart.pprint()
# Use this grammar for the rest of the assignment
grammar2=parse_grammar([
"S -> Sdecl '.' | Simp '.' | Sq '?' ",
"Sdecl -> NP VP",
"Simp -> VP",
"Sq -> Sqyn | Swhadv",
"Sqyn -> Mod Sdecl | Aux Sdecl",
"Swhadv -> WhAdv Sqyn",
"Sc -> Subconj Sdecl",
"NP -> PropN | Pro | NP0 ",
"NP0 -> NP1 | NP0 PP",
"Prep -> 'to' | 'with'",
"Vi -> 'ate'", #intransitive
"Vt -> 'ate' | 'book' | 'Book' | 'gave' | 'told'", #transitive
"Vdt -> 'gave' | 'told' ", #ditransitive
"Subconj -> 'that'", #subordinating conjunction
"Mod -> 'Can' | 'will'", #modal verbs
"Aux -> 'did' ", #auxiliary verbs
"WhAdv -> 'Why'",
"PropN -> 'John' | 'Mary' | 'NYC' | 'London'",
"Adj -> 'nice' | 'drawing'",
"Pro -> 'you' | 'he'",
"Adv -> 'today'"
])
print(grammar) #the simpler grammar
chart = CKY(grammar)
#this illustrates tracing of a very simple sentence; feel free to try others.
# chart.recognise(tokenise("the frogs swim"),True)
# chart.pprint()
#build a chart with the larger grammar
chart2 = CKY(grammar2)
# Note, please do _not_ use the Tree.draw() method uncommented
# _anywhere in this file_ (you are encouraged to use it in preparing
# your report).
# The sentences to examine.
#
for s in ["Book a flight to NYC."]:
print(s, chart2.recognise(tokenise(s), True))