def java_parse_sentence(self, sequence: str):
json_grammar = self.grammar.save_to_json()
gateway = JavaGateway()
cyk = gateway.entry_point
cyk_result = cyk.runCyk(sequence, json_grammar)
result = json.loads(cyk_result)
rt = result['rules_table']
pt = result['probability_table']
for i in range(len(sequence)):
for j in range(len(sequence) - i):
rules = []
pts = []
for rule in rt[i][j]['rules']:
rules.append(sCellRule.fromCellRuleDict(rule))
rt[i][j] = rules
for p in pt[i][j]:
if p['item_1'] != -1 and p['item_2'] != -1:
pts.append(
ProbabilityArrayCell(p['item_1'], p['item_2']))
else:
pts.append(self.default_value)
pt[i][j] = pts
self.rules_table = rt
self.probability_array = pt
if self.learning_on:
rules_to_add = result['rules_to_add']
for cell_indexes in rules_to_add:
self.__apply_aggressive_and_final_covering(
cell_indexes['i'], cell_indexes['j'])
def __init_cell(cyk_values: CykValues, index: int, rule: sRule):
rule_left_index = rule.left.index
cyk_values.probability_array[0][index][rule_left_index] = ProbabilityArrayCell()
if cyk_values.mode == "Viterbi":
cyk_values.probability_array[0][index][rule_left_index].item_1 = math.log10(rule.probability)
else:
cyk_values.probability_array[0][index][rule_left_index].item_1 = rule.probability
cyk_values.probability_array[0][index][rule_left_index].item_2 = rule.probability
def __init_cell(cyk_values: CykValues, index: int, rule: sRule):
rule_left_index = rule.left.index
cyk_values.probability_array[0][index][
rule_left_index] = ProbabilityArrayCell()
cyk_values.probability_array[0][index][
rule_left_index].item_1 = rule.probability
cyk_values.probability_array[0][index][
rule_left_index].item_2 = rule.probability
def __init_cell(self, index: int, rule: sRule):
rule_left_index = rule.left.index
self.probability_array[0][index][
rule_left_index] = ProbabilityArrayCell()
if self.mode == "Viterbi":
self.probability_array[0][index][
rule_left_index].item_1 = math.log10(rule.probability)
else:
self.probability_array[0][index][
rule_left_index].item_1 = rule.probability
self.probability_array[0][index][
rule_left_index].item_2 = rule.probability
def __calculate_min_prob_rule_cell_probability(default_value, cell_probability: ProbabilityArrayCell,
parent_cell_1_probability: ProbabilityArrayCell,
parent_cell_2_probability: ProbabilityArrayCell,
rule: sRule):
if cell_probability == default_value:
cell_probability = ProbabilityArrayCell()
cell_probability.item_1 = min(min(rule.probability * parent_cell_1_probability.item_1),
parent_cell_2_probability.item_1)
cell_probability.item_2 = rule.probability * \
parent_cell_1_probability.item_2 * \
parent_cell_2_probability.item_2
else:
cell_probability.item_1 = max(cell_probability.item_1,
min(min(rule.probability), parent_cell_1_probability.item_1,
parent_cell_2_probability.item_1))
cell_probability.item_2 = cell_probability.item_2 + rule.probability * \
parent_cell_1_probability.item_2 * \
parent_cell_2_probability.item_2
return cell_probability
def __calculate_viterbi_rule_cell_probability(default_value, cell_probability: ProbabilityArrayCell,
parent_cell_1_probability: ProbabilityArrayCell,
parent_cell_2_probability: ProbabilityArrayCell,
rule: sRule):
if cell_probability == default_value:
cell_probability = ProbabilityArrayCell()
cell_probability.item_1 = math.log10(rule.probability) + \
parent_cell_1_probability.item_1 * parent_cell_2_probability.item_1
cell_probability.item_2 = (rule.probability *
parent_cell_1_probability.item_2 *
parent_cell_2_probability.item_2)
else:
cell_probability.item_1 = max(cell_probability.item_1, math.log10(parent_cell_1_probability.item_1 +
parent_cell_2_probability.item_1))
cell_probability.item_2 = cell_probability.item_2 + (rule.probability *
parent_cell_1_probability.item_2 *
parent_cell_2_probability.item_2)
return cell_probability
def __calculate_baum_welch_rule_cell_probability(default_value, cell_probability: ProbabilityArrayCell,
parent_cell_1_probability: ProbabilityArrayCell,
parent_cell_2_probability: ProbabilityArrayCell,
rule: sRule) -> ProbabilityArrayCell:
if cell_probability == default_value:
cell_probability = ProbabilityArrayCell()
cell_probability.item_1 = rule.probability * \
parent_cell_1_probability.item_1 * \
parent_cell_2_probability.item_1
cell_probability.item_2 = rule.probability * \
parent_cell_1_probability.item_2 * \
parent_cell_2_probability.item_2
else:
cell_probability.item_1 = cell_probability.item_1 + \
(rule.probability *
parent_cell_1_probability.item_1 *
parent_cell_2_probability.item_1)
cell_probability.item_2 = cell_probability.item_2 + \
(rule.probability *
parent_cell_1_probability.item_2 *
parent_cell_2_probability.item_2)
return cell_probability