def test_training_data4(self):
from experiment import sample_wins_and_losses
from metrics import SemanticsOracleAccuracyMetric
from scoring import Model
from travel import TravelDomain
from geonames import GeoNamesAnnotator
domain = TravelDomain()
rules = self.rules_travel + self.rules_travel_locations + self.rules_travel_modes + self.rules_travel_triggers + self.rules_request_types + self.rules_optionals
grammar = Unit2Grammar(rules=rules, annotators=[GeoNamesAnnotator(live_requests=False)])
model = Model(grammar=grammar)
metric = SemanticsOracleAccuracyMetric()
# If printing=True, prints a sampling of wins (correct semantics in
# first parse) and losses on the dataset.
metric_values = sample_wins_and_losses(domain=domain, model=model, metric=metric, seed=31, printing=False)
def travel_demo():
from travel import TravelDomain
demo_learning_from_semantics(TravelDomain())
# Extend domain.ops:
# Make sure things are working:
display_examples(('the one', 'the average of one and four'),
grammar=math_grammar,
domain=math_domain)
# ## Travel domain
# Here's an illustration of how parsing and interpretation work in this domain:
# In[ ]:
from travel import TravelDomain
travel_domain = TravelDomain()
travel_grammar = travel_domain.grammar()
display_examples(("flight from Boston to San Francisco",
"directions from New York to Philadelphia",
"directions New York to Philadelphia"),
grammar=travel_grammar,
domain=travel_domain)
# For these questions, we'll combine grammars with machine learning.
# Here's now to train and evaluate a model using the grammar
# that is included in `TravelDomain` along with a basic feature
# function.
# In[ ]: