spaCy/spacy/syntax/arc_eager.pyx

631 lines
22 KiB
Cython
Raw Normal View History

# cython: profile=True
# cython: cdivision=True
# cython: infer_types=True
2017-04-15 14:05:15 +03:00
# coding: utf-8
from __future__ import unicode_literals
2017-10-27 20:45:57 +03:00
from cpython.ref cimport Py_INCREF
2015-06-09 02:41:09 +03:00
from cymem.cymem cimport Pool
from collections import OrderedDict, defaultdict, Counter
from thinc.extra.search cimport Beam
import json
2017-04-15 14:05:15 +03:00
2018-04-29 16:49:11 +03:00
from .nonproj import is_nonproj_tree
from ..typedefs cimport hash_t, attr_t
from ..strings cimport hash_string
from .stateclass cimport StateClass
2017-10-27 20:45:57 +03:00
from ._state cimport StateC
from . import nonproj
2017-04-15 14:05:15 +03:00
from .transition_system cimport move_cost_func_t, label_cost_func_t
2017-10-27 20:45:57 +03:00
from ..gold cimport GoldParse, GoldParseC
2017-04-15 14:05:15 +03:00
from ..structs cimport TokenC
from ..errors import Errors
2018-05-02 17:19:22 +03:00
from ..tokens.doc cimport Doc, set_children_from_heads
2015-06-09 02:41:09 +03:00
# Calculate cost as gold/not gold. We don't use scalar value anyway.
cdef int BINARY_COSTS = 1
cdef weight_t MIN_SCORE = -90000
cdef attr_t SUBTOK_LABEL = hash_string('subtok')
DEF NON_MONOTONIC = True
DEF USE_BREAK = True
2014-12-16 14:44:43 +03:00
# Break transition from here
# http://www.aclweb.org/anthology/P13-1074
2014-12-16 14:44:43 +03:00
cdef enum:
SHIFT
REDUCE
LEFT
RIGHT
2015-05-11 17:12:03 +03:00
BREAK
2015-05-11 17:12:03 +03:00
2014-12-16 14:44:43 +03:00
N_MOVES
2015-05-11 17:12:03 +03:00
MOVE_NAMES = [None] * N_MOVES
MOVE_NAMES[SHIFT] = 'S'
MOVE_NAMES[REDUCE] = 'D'
MOVE_NAMES[LEFT] = 'L'
MOVE_NAMES[RIGHT] = 'R'
MOVE_NAMES[BREAK] = 'B'
2015-01-27 19:18:43 +03:00
# Helper functions for the arc-eager oracle
cdef weight_t push_cost(StateClass stcls, const GoldParseC* gold, int target) nogil:
cdef weight_t cost = 0
cdef int i, S_i
for i in range(stcls.stack_depth()):
S_i = stcls.S(i)
if gold.heads[target] == S_i:
cost += 1
if gold.heads[S_i] == target and (NON_MONOTONIC or not stcls.has_head(S_i)):
cost += 1
if BINARY_COSTS and cost >= 1:
return cost
2017-05-30 21:37:24 +03:00
cost += Break.is_valid(stcls.c, 0) and Break.move_cost(stcls, gold) == 0
return cost
cdef weight_t pop_cost(StateClass stcls, const GoldParseC* gold, int target) nogil:
cdef weight_t cost = 0
cdef int i, B_i
for i in range(stcls.buffer_length()):
B_i = stcls.B(i)
cost += gold.heads[B_i] == target
cost += gold.heads[target] == B_i
if gold.heads[B_i] == B_i or gold.heads[B_i] < target:
break
if BINARY_COSTS and cost >= 1:
return cost
2017-05-30 21:37:24 +03:00
if Break.is_valid(stcls.c, 0) and Break.move_cost(stcls, gold) == 0:
cost += 1
return cost
cdef weight_t arc_cost(StateClass stcls, const GoldParseC* gold, int head, int child) nogil:
if arc_is_gold(gold, head, child):
return 0
elif stcls.H(child) == gold.heads[child]:
return 1
# Head in buffer
2017-05-30 21:37:24 +03:00
elif gold.heads[child] >= stcls.B(0) and stcls.B(1) != 0:
return 1
else:
return 0
2015-06-10 07:56:35 +03:00
cdef bint arc_is_gold(const GoldParseC* gold, int head, int child) nogil:
2017-05-30 21:37:24 +03:00
if not gold.has_dep[child]:
return True
elif gold.heads[child] == head:
return True
else:
return False
2017-05-28 15:06:40 +03:00
cdef bint label_is_gold(const GoldParseC* gold, int head, int child, attr_t label) nogil:
2017-05-30 21:37:24 +03:00
if not gold.has_dep[child]:
return True
2017-05-30 21:37:24 +03:00
elif label == 0:
return True
elif gold.labels[child] == label:
return True
else:
return False
2015-06-10 07:56:35 +03:00
cdef bint _is_gold_root(const GoldParseC* gold, int word) nogil:
2017-05-30 21:37:24 +03:00
return gold.heads[word] == word or not gold.has_dep[word]
cdef class Shift:
@staticmethod
2017-05-28 15:06:40 +03:00
cdef bint is_valid(const StateC* st, attr_t label) nogil:
sent_start = st._sent[st.B_(0).l_edge].sent_start
return st.buffer_length() >= 2 and not st.shifted[st.B(0)] and sent_start != 1
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef int transition(StateC* st, attr_t label) nogil:
st.push()
st.fast_forward()
2014-12-16 14:44:43 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef weight_t cost(StateClass st, const GoldParseC* gold, attr_t label) nogil:
return Shift.move_cost(st, gold) + Shift.label_cost(st, gold, label)
@staticmethod
cdef inline weight_t move_cost(StateClass s, const GoldParseC* gold) nogil:
return push_cost(s, gold, s.B(0))
@staticmethod
2017-05-28 15:06:40 +03:00
cdef inline weight_t label_cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return 0
cdef class Reduce:
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef bint is_valid(const StateC* st, attr_t label) nogil:
return st.stack_depth() >= 2
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef int transition(StateC* st, attr_t label) nogil:
if st.has_head(st.S(0)):
st.pop()
else:
st.unshift()
st.fast_forward()
@staticmethod
2017-05-28 15:06:40 +03:00
cdef weight_t cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return Reduce.move_cost(s, gold) + Reduce.label_cost(s, gold, label)
@staticmethod
cdef inline weight_t move_cost(StateClass st, const GoldParseC* gold) nogil:
cost = pop_cost(st, gold, st.S(0))
if not st.has_head(st.S(0)):
# Decrement cost for the arcs e save
for i in range(1, st.stack_depth()):
S_i = st.S(i)
if gold.heads[st.S(0)] == S_i:
cost -= 1
if gold.heads[S_i] == st.S(0):
cost -= 1
2017-05-30 21:37:24 +03:00
if Break.is_valid(st.c, 0) and Break.move_cost(st, gold) == 0:
cost -= 1
return cost
@staticmethod
2017-05-28 15:06:40 +03:00
cdef inline weight_t label_cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return 0
cdef class LeftArc:
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef bint is_valid(const StateC* st, attr_t label) nogil:
if label == SUBTOK_LABEL and st.S(0) != (st.B(0)-1):
2018-04-08 15:50:21 +03:00
return 0
sent_start = st._sent[st.B_(0).l_edge].sent_start
return sent_start != 1
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef int transition(StateC* st, attr_t label) nogil:
st.add_arc(st.B(0), st.S(0), label)
st.pop()
st.fast_forward()
@staticmethod
2017-05-28 15:06:40 +03:00
cdef weight_t cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return LeftArc.move_cost(s, gold) + LeftArc.label_cost(s, gold, label)
@staticmethod
cdef inline weight_t move_cost(StateClass s, const GoldParseC* gold) nogil:
cdef weight_t cost = 0
if arc_is_gold(gold, s.B(0), s.S(0)):
# Have a negative cost if we 'recover' from the wrong dependency
return 0 if not s.has_head(s.S(0)) else -1
else:
# Account for deps we might lose between S0 and stack
if not s.has_head(s.S(0)):
for i in range(1, s.stack_depth()):
cost += gold.heads[s.S(i)] == s.S(0)
cost += gold.heads[s.S(0)] == s.S(i)
return cost + pop_cost(s, gold, s.S(0)) + arc_cost(s, gold, s.B(0), s.S(0))
2014-12-16 14:44:43 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef inline weight_t label_cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return arc_is_gold(gold, s.B(0), s.S(0)) and not label_is_gold(gold, s.B(0), s.S(0), label)
cdef class RightArc:
@staticmethod
2017-05-28 15:06:40 +03:00
cdef bint is_valid(const StateC* st, attr_t label) nogil:
# If there's (perhaps partial) parse pre-set, don't allow cycle.
if label == SUBTOK_LABEL and st.S(0) != (st.B(0)-1):
2018-04-08 15:50:21 +03:00
return 0
sent_start = st._sent[st.B_(0).l_edge].sent_start
return sent_start != 1 and st.H(st.S(0)) != st.B(0)
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef int transition(StateC* st, attr_t label) nogil:
st.add_arc(st.S(0), st.B(0), label)
st.push()
st.fast_forward()
@staticmethod
2017-05-28 15:06:40 +03:00
cdef inline weight_t cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return RightArc.move_cost(s, gold) + RightArc.label_cost(s, gold, label)
@staticmethod
cdef inline weight_t move_cost(StateClass s, const GoldParseC* gold) nogil:
if arc_is_gold(gold, s.S(0), s.B(0)):
return 0
2016-02-01 04:37:08 +03:00
elif s.c.shifted[s.B(0)]:
return push_cost(s, gold, s.B(0))
else:
return push_cost(s, gold, s.B(0)) + arc_cost(s, gold, s.S(0), s.B(0))
@staticmethod
2017-05-28 15:06:40 +03:00
cdef weight_t label_cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return arc_is_gold(gold, s.S(0), s.B(0)) and not label_is_gold(gold, s.S(0), s.B(0), label)
cdef class Break:
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef bint is_valid(const StateC* st, attr_t label) nogil:
2015-06-09 02:41:09 +03:00
cdef int i
if not USE_BREAK:
return False
elif st.at_break():
2015-06-09 02:41:09 +03:00
return False
elif st.stack_depth() < 1:
return False
elif st.B_(0).l_edge < 0:
return False
2017-10-09 01:02:45 +03:00
elif st._sent[st.B_(0).l_edge].sent_start < 0:
return False
2015-06-09 02:41:09 +03:00
else:
2017-08-26 03:50:55 +03:00
return True
2015-06-09 02:41:09 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef int transition(StateC* st, attr_t label) nogil:
st.set_break(st.B_(0).l_edge)
st.fast_forward()
@staticmethod
2017-05-28 15:06:40 +03:00
cdef weight_t cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return Break.move_cost(s, gold) + Break.label_cost(s, gold, label)
@staticmethod
cdef inline weight_t move_cost(StateClass s, const GoldParseC* gold) nogil:
cdef weight_t cost = 0
cdef int i, j, S_i, B_i
for i in range(s.stack_depth()):
S_i = s.S(i)
for j in range(s.buffer_length()):
B_i = s.B(j)
cost += gold.heads[S_i] == B_i
cost += gold.heads[B_i] == S_i
if cost != 0:
return cost
# Check for sentence boundary --- if it's here, we can't have any deps
# between stack and buffer, so rest of action is irrelevant.
s0_root = _get_root(s.S(0), gold)
b0_root = _get_root(s.B(0), gold)
if s0_root != b0_root or s0_root == -1 or b0_root == -1:
return cost
else:
return cost + 1
2015-08-09 00:31:54 +03:00
@staticmethod
2017-05-28 15:06:40 +03:00
cdef inline weight_t label_cost(StateClass s, const GoldParseC* gold, attr_t label) nogil:
return 0
cdef int _get_root(int word, const GoldParseC* gold) nogil:
2017-08-26 03:50:55 +03:00
while gold.heads[word] != word and gold.has_dep[word] and word >= 0:
word = gold.heads[word]
2017-05-30 21:37:24 +03:00
if not gold.has_dep[word]:
return -1
else:
return word
2015-08-09 00:31:54 +03:00
2017-03-11 20:12:01 +03:00
cdef void* _init_state(Pool mem, int length, void* tokens) except NULL:
2017-11-14 04:11:40 +03:00
st = new StateC(<const TokenC*>tokens, length)
for i in range(st.length):
if st._sent[i].dep == 0:
st._sent[i].l_edge = i
st._sent[i].r_edge = i
st._sent[i].head = 0
st._sent[i].dep = 0
st._sent[i].l_kids = 0
st._sent[i].r_kids = 0
2017-03-11 20:12:01 +03:00
st.fast_forward()
return <void*>st
cdef class ArcEager(TransitionSystem):
2017-03-11 20:12:01 +03:00
def __init__(self, *args, **kwargs):
TransitionSystem.__init__(self, *args, **kwargs)
self.init_beam_state = _init_state
@classmethod
def get_actions(cls, **kwargs):
min_freq = kwargs.get('min_freq', None)
actions = defaultdict(lambda: Counter())
actions[SHIFT][''] = 1
actions[REDUCE][''] = 1
for label in kwargs.get('left_labels', []):
actions[LEFT][label] = 1
actions[SHIFT][label] = 1
for label in kwargs.get('right_labels', []):
actions[RIGHT][label] = 1
actions[REDUCE][label] = 1
for example in kwargs.get('gold_parses', []):
for token_annotation in example.token_annotations:
heads, labels = nonproj.projectivize(token_annotation.heads, token_annotation.deps)
for child, head, label in zip(token_annotation.ids, heads, labels):
if label.upper() == 'ROOT' :
label = 'ROOT'
if head == child:
actions[BREAK][label] += 1
elif head < child:
actions[RIGHT][label] += 1
actions[REDUCE][''] += 1
elif head > child:
actions[LEFT][label] += 1
actions[SHIFT][''] += 1
if min_freq is not None:
for action, label_freqs in actions.items():
for label, freq in list(label_freqs.items()):
if freq < min_freq:
label_freqs.pop(label)
# Ensure these actions are present
actions[BREAK].setdefault('ROOT', 0)
if kwargs.get("learn_tokens") is True:
actions[RIGHT].setdefault('subtok', 0)
actions[LEFT].setdefault('subtok', 0)
# Used for backoff
actions[RIGHT].setdefault('dep', 0)
actions[LEFT].setdefault('dep', 0)
return actions
@property
def action_types(self):
return (SHIFT, REDUCE, LEFT, RIGHT, BREAK)
def get_cost(self, StateClass state, GoldParse gold, action):
cdef Transition t = self.lookup_transition(action)
if not t.is_valid(state.c, t.label):
return 9000
else:
return t.get_cost(state, &gold.c, t.label)
def transition(self, StateClass state, action):
cdef Transition t = self.lookup_transition(action)
t.do(state.c, t.label)
return state
def is_gold_parse(self, StateClass state, GoldParse gold):
predicted = set()
truth = set()
for i in range(gold.length):
if gold.cand_to_gold[i] is None:
continue
if state.safe_get(i).dep:
2017-10-27 20:45:57 +03:00
predicted.add((i, state.H(i),
self.strings[state.safe_get(i).dep]))
else:
predicted.add((i, state.H(i), 'ROOT'))
id_ = gold.orig.ids[gold.cand_to_gold[i]]
head = gold.orig.heads[gold.cand_to_gold[i]]
dep = gold.orig.deps[gold.cand_to_gold[i]]
truth.add((id_, head, dep))
return truth == predicted
2017-05-26 19:31:23 +03:00
def has_gold(self, GoldParse gold, start=0, end=None):
end = end or len(gold.heads)
if all([tag is None for tag in gold.heads[start:end]]):
return False
else:
return True
def preprocess_gold(self, GoldParse gold):
2017-05-26 19:31:23 +03:00
if not self.has_gold(gold):
return None
# Figure out whether we're using subtok
use_subtok = False
for action, labels in self.labels.items():
if SUBTOK_LABEL in labels:
use_subtok = True
break
for i, (head, dep) in enumerate(zip(gold.heads, gold.labels)):
2017-10-27 20:45:57 +03:00
# Missing values
if head is None or dep is None:
gold.c.heads[i] = i
2017-05-30 21:37:24 +03:00
gold.c.has_dep[i] = False
elif dep == SUBTOK_LABEL and not use_subtok:
# If we're not doing the joint tokenization and parsing,
# regard these subtok labels as missing
gold.c.heads[i] = i
gold.c.labels[i] = 0
gold.c.has_dep[i] = False
else:
if head > i:
action = LEFT
elif head < i:
action = RIGHT
else:
action = BREAK
if dep not in self.labels[action]:
if action == BREAK:
dep = 'ROOT'
elif nonproj.is_decorated(dep):
backoff = nonproj.decompose(dep)[0]
if backoff in self.labels[action]:
dep = backoff
else:
dep = 'dep'
else:
dep = 'dep'
2017-05-30 21:37:24 +03:00
gold.c.has_dep[i] = True
if dep.upper() == 'ROOT':
dep = 'ROOT'
gold.c.heads[i] = head
gold.c.labels[i] = self.strings.add(dep)
return gold
def get_beam_parses(self, Beam beam):
parses = []
probs = beam.probs
for i in range(beam.size):
state = <StateC*>beam.at(i)
if state.is_final():
self.finalize_state(state)
prob = probs[i]
parse = []
for j in range(state.length):
head = state.H(j)
label = self.strings[state._sent[j].dep]
parse.append((head, j, label))
parses.append((prob, parse))
return parses
cdef Transition lookup_transition(self, object name_or_id) except *:
if isinstance(name_or_id, int):
return self.c[name_or_id]
name = name_or_id
if '-' in name:
move_str, label_str = name.split('-', 1)
2015-08-09 00:31:54 +03:00
label = self.strings[label_str]
else:
2015-08-09 00:31:54 +03:00
move_str = name
label = 0
move = MOVE_NAMES.index(move_str)
for i in range(self.n_moves):
if self.c[i].move == move and self.c[i].label == label:
return self.c[i]
return Transition(clas=0, move=MISSING, label=0)
2017-05-28 15:06:40 +03:00
def move_name(self, int move, attr_t label):
label_str = self.strings[label]
if label_str:
return MOVE_NAMES[move] + '-' + label_str
else:
return MOVE_NAMES[move]
def class_name(self, int i):
return self.move_name(self.c[i].move, self.c[i].label)
2017-05-28 15:06:40 +03:00
cdef Transition init_transition(self, int clas, int move, attr_t label) except *:
# TODO: Apparent Cython bug here when we try to use the Transition()
# constructor with the function pointers
cdef Transition t
t.score = 0
t.clas = clas
t.move = move
t.label = label
if move == SHIFT:
t.is_valid = Shift.is_valid
t.do = Shift.transition
t.get_cost = Shift.cost
elif move == REDUCE:
t.is_valid = Reduce.is_valid
t.do = Reduce.transition
t.get_cost = Reduce.cost
elif move == LEFT:
t.is_valid = LeftArc.is_valid
t.do = LeftArc.transition
t.get_cost = LeftArc.cost
elif move == RIGHT:
t.is_valid = RightArc.is_valid
t.do = RightArc.transition
t.get_cost = RightArc.cost
elif move == BREAK:
t.is_valid = Break.is_valid
t.do = Break.transition
t.get_cost = Break.cost
else:
raise ValueError(Errors.E019.format(action=move, src='arc_eager'))
return t
cdef int initialize_state(self, StateC* st) nogil:
for i in range(st.length):
if st._sent[i].dep == 0:
st._sent[i].l_edge = i
st._sent[i].r_edge = i
st._sent[i].head = 0
st._sent[i].dep = 0
st._sent[i].l_kids = 0
st._sent[i].r_kids = 0
st.fast_forward()
cdef int finalize_state(self, StateC* st) nogil:
cdef int i
for i in range(st.length):
if st._sent[i].head == 0:
st._sent[i].dep = self.root_label
2018-05-02 17:19:22 +03:00
def finalize_doc(self, Doc doc):
doc.is_parsed = True
2018-05-02 17:19:22 +03:00
set_children_from_heads(doc.c, doc.length)
2016-02-01 05:07:37 +03:00
cdef int set_valid(self, int* output, const StateC* st) nogil:
cdef bint[N_MOVES] is_valid
2017-05-30 21:37:24 +03:00
is_valid[SHIFT] = Shift.is_valid(st, 0)
is_valid[REDUCE] = Reduce.is_valid(st, 0)
is_valid[LEFT] = LeftArc.is_valid(st, 0)
is_valid[RIGHT] = RightArc.is_valid(st, 0)
is_valid[BREAK] = Break.is_valid(st, 0)
cdef int i
for i in range(self.n_moves):
if self.c[i].label == SUBTOK_LABEL:
output[i] = self.c[i].is_valid(st, self.c[i].label)
else:
output[i] = is_valid[self.c[i].move]
cdef int set_costs(self, int* is_valid, weight_t* costs,
StateClass stcls, GoldParse gold) except -1:
2017-05-30 21:37:24 +03:00
cdef int i, move
cdef attr_t label
cdef label_cost_func_t[N_MOVES] label_cost_funcs
cdef move_cost_func_t[N_MOVES] move_cost_funcs
cdef weight_t[N_MOVES] move_costs
for i in range(N_MOVES):
move_costs[i] = 9000
move_cost_funcs[SHIFT] = Shift.move_cost
move_cost_funcs[REDUCE] = Reduce.move_cost
move_cost_funcs[LEFT] = LeftArc.move_cost
move_cost_funcs[RIGHT] = RightArc.move_cost
move_cost_funcs[BREAK] = Break.move_cost
label_cost_funcs[SHIFT] = Shift.label_cost
label_cost_funcs[REDUCE] = Reduce.label_cost
label_cost_funcs[LEFT] = LeftArc.label_cost
label_cost_funcs[RIGHT] = RightArc.label_cost
label_cost_funcs[BREAK] = Break.label_cost
2017-05-28 15:06:40 +03:00
cdef attr_t* labels = gold.c.labels
cdef int* heads = gold.c.heads
n_gold = 0
for i in range(self.n_moves):
if self.c[i].is_valid(stcls.c, self.c[i].label):
is_valid[i] = True
move = self.c[i].move
label = self.c[i].label
if move_costs[move] == 9000:
move_costs[move] = move_cost_funcs[move](stcls, &gold.c)
costs[i] = move_costs[move] + label_cost_funcs[move](stcls, &gold.c, label)
n_gold += costs[i] <= 0
2015-06-10 07:56:35 +03:00
else:
is_valid[i] = False
costs[i] = 9000
if n_gold < 1:
# Check projectivity --- leading cause
if is_nonproj_tree(gold.heads):
raise ValueError(Errors.E020)
else:
failure_state = stcls.print_state(gold.words)
raise ValueError(Errors.E021.format(n_actions=self.n_moves,
state=failure_state))
def get_beam_annot(self, Beam beam):
2017-11-14 04:11:40 +03:00
length = (<StateC*>beam.at(0)).length
heads = [{} for _ in range(length)]
deps = [{} for _ in range(length)]
probs = beam.probs
for i in range(beam.size):
2017-11-14 04:11:40 +03:00
state = <StateC*>beam.at(i)
self.finalize_state(state)
if state.is_final():
prob = probs[i]
2017-11-14 04:11:40 +03:00
for j in range(state.length):
head = j + state._sent[j].head
dep = state._sent[j].dep
heads[j].setdefault(head, 0.0)
heads[j][head] += prob
deps[j].setdefault(dep, 0.0)
deps[j][dep] += prob
return heads, deps