def do_rule(sentdict, aux, tree, word_positions_in_tree):
auxidx = aux.wordnum
try:
if sentdict.lemmas[auxidx+1] == 'that':
return True
except IndexError: pass
if not aux_locally_ccommanded_by_verb(sentdict, aux, tree, word_positions_in_tree):
if to_precedes_aux(sentdict, aux): return False
localt = nt.generate_local_structure_from_subtree(tree, tree[word_positions_in_tree[auxidx-1]])
local_word_subtrees = nt.getsmallestsubtrees(localt)
try:
checkpuncttag = sentdict.pos[auxidx+1]
if is_period(checkpuncttag) or is_comma(checkpuncttag) or is_dash_or_colon(checkpuncttag):
endbool = True
for subtree in local_word_subtrees:
if is_verb(subtree.label()) and subtree != tree[word_positions_in_tree[auxidx-1]]:
if nt.ccommands(localt, subtree, tree[word_positions_in_tree[auxidx-1]]):
endbool = False
break
if endbool:
return endbool
except IndexError:
pass
# Don't at the end of sentence.
try:
checkpuncttag = sentdict.pos[auxidx+2]
if sentdict.lemmas[auxidx+1] == 'not' and (is_period(checkpuncttag) or is_comma(checkpuncttag) or is_dash_or_colon(checkpuncttag)):
endbool = True
for subtree in local_word_subtrees:
if is_verb(subtree.label()):
if nt.ccommands(localt, subtree, tree[word_positions_in_tree[auxidx-1]]):
endbool = False
break
if endbool:
return endbool
except IndexError:
pass
if is_ccommanded_by_continuation_word(sentdict ,aux, tree, word_positions_in_tree):
return True
if verb_follows_aux(sentdict, aux):
return False
try:
if is_preposition(sentdict.pos[auxidx+1]) and sentdict.words[auxidx] != 'done':
return True
except IndexError:
pass
return False
def auxlocallyccommandsverb(sentdict, auxidx, t, word_positions_in_tree):
try:
localt = nltktree.generate_local_structure_from_subtree(t, t[word_positions_in_tree[auxidx-1]])
local_word_subtrees = nltktree.getsmallestsubtrees(localt)
for subtree in local_word_subtrees:
if isverb(subtree.label()):
if nltktree.ccommands(localt, t[word_positions_in_tree[auxidx-1]], subtree)\
and not nltktree.ccommands(localt, subtree, t[word_positions_in_tree[auxidx-1]]):
return True
except IndexError: pass
return False
def aux_locally_ccommands_verb(sentdict, aux, tree, word_positions_in_tree):
try:
localt = nt.generate_local_structure_from_subtree(tree, tree[word_positions_in_tree[aux.wordnum-1]])
local_word_subtrees = nt.getsmallestsubtrees(localt)
for subtree in local_word_subtrees:
if is_verb(subtree.label()):
if nt.ccommands(localt, tree[word_positions_in_tree[aux.wordnum-1]], subtree)\
and not nt.ccommands(localt, subtree, tree[word_positions_in_tree[aux.wordnum-1]]):
return True
except IndexError: pass
return False
def is_ccommanded_by_continuation_word(sentdict, aux, tree, word_positions_in_tree):
for i in range(max(0,aux.wordnum-CONTINUATION_WORD_DISTANCE_SEARCH),aux.wordnum):
crt_word = sentdict.words[i].lower()
if crt_word in ['than','as','so']:
if nt.ccommands(tree, tree[word_positions_in_tree[i-1]], tree[word_positions_in_tree[aux.wordnum-1]]):
return True
return False
def auxccommandsverb(sentdict, auxidx, t, word_positions_in_tree):
subtrees = nltktree.getsmallestsubtrees(t)
for subtree in subtrees:
if isverb(subtree.label()):
try:
if nltktree.ccommands(t, t[word_positions_in_tree[auxidx-1]], subtree):
return True
except IndexError:
pass
return False
def aux_ccommanded_by_verb(sentdict, aux, tree, word_positions_in_tree):
subtrees = nt.getsmallestsubtrees(tree)
for subtree in subtrees:
if is_verb(subtree.label()):
try:
if nt.ccommands(tree, subtree, tree[word_positions_in_tree[aux.wordnum-1]]):
return True
except IndexError:
pass
return False
def isccommandedbycontinuationword(sentdict, auxidx, t, word_positions_in_tree):
continuation_words = ['than', 'as', 'so']
local_continuation_words = ['like']
for i in range(0,len(sentdict['words'])):
crtword = sentdict['words'][i].lower()
if crtword in continuation_words and i >= auxidx - 5 and i < auxidx:
if nltktree.ccommands(t, t[word_positions_in_tree[i-1]], t[word_positions_in_tree[auxidx-1]]):
return True
# localt = nltktree.generatelocalstructurefromsubtree(t, t[word_positions_in_tree[auxidx-1]])
# local_word_subtrees = nltktree.getsmallestsubtrees(localt)
# Checking for local c-command.
# for subtree in local_word_subtrees:
# if subtree.leaves() in local_continuation_words:
# if nltktree.ccommands(localt, subtree, t[word_positions_in_tree[auxidx-1]]):
# return True
return False
def docheck(sentdict, auxidx, t, word_positions_in_tree, verbose=False):
# We DO NOT want to consider 'do so' or 'do the same' sentences here!
"""try:
if sentdict['lemmas'][auxidx+1] == 'so' or (sentdict['lemmas'][auxidx+1] == 'the' and sentdict['lemmas'][auxidx+2] == 'same'):
return False
# We can be POSITIVE that there is NO vpe if we have 'don't do ...' or 'x does do ...'
if sentdict['lemmas'][auxidx-1] == 'do' or (sentdict['lemmas'][auxidx-2] == 'do' and sentdict['lemmas'][auxidx-1] == 'not'):
return False
if sentdict['lemmas'][auxidx+1] == 'do' or (sentdict['lemmas'][auxidx+1] == 'do' and sentdict['lemmas'][auxidx+2] == 'not'):
return False
except IndexError: pass"""
try:
if sentdict['lemmas'][auxidx+1] == 'that':
return True
except IndexError: pass
if not auxislocallyccommandedbyverb(sentdict, auxidx, t, word_positions_in_tree):
# # If 'do' locally c-commands a verb AND is locally c-commanded by a verb, we can be basically 100% sure that there is no VPE.
# if auxislocallyccommandedbyverb(sentdict, auxidx, tree, word_positions_in_tree) and auxlocallyccommandsverb(sentdict, auxidx, tree, word_positions_in_tree):
# return False
if toprecedesaux(sentdict, auxidx): return False
localt = nltktree.generate_local_structure_from_subtree(t, t[word_positions_in_tree[auxidx-1]])
local_word_subtrees = nltktree.getsmallestsubtrees(localt)
# Do at the end of sentence.
checkpuncttag = sentdict['pos'][auxidx+1]
if isperiod(checkpuncttag) or iscomma(checkpuncttag) or isdashorcolon(checkpuncttag):
endbool = True
for subtree in local_word_subtrees:
if isverb( subtree.label() ) and subtree != t[word_positions_in_tree[auxidx-1]]:
if nltktree.ccommands(localt, subtree, t[word_positions_in_tree[auxidx-1]]):
endbool = False
break
if endbool: return endbool
# Don't at the end of sentence.
try:
checkpuncttag = sentdict['pos'][auxidx+2]
if sentdict['lemmas'][auxidx+1] == 'not' and (isperiod(checkpuncttag) or iscomma(checkpuncttag) or isdashorcolon(checkpuncttag)):
endbool = True
for subtree in local_word_subtrees:
if isverb( subtree.label() ):
if nltktree.ccommands(localt, subtree, t[word_positions_in_tree[auxidx-1]]):
endbool = False
break
if endbool: return endbool
except IndexError:
pass
# Small increase in recall, decrease in precision from this.
# numverbs = 0
# for subtree in local_word_subtrees:
# if isverb(subtree.label()) or isnounorprep(subtree.label()):
# numverbs+=1
# if numverbs == 1:
# return True
if isccommandedbycontinuationword(sentdict ,auxidx, t, word_positions_in_tree):
#if not auxlocallyccommandsverb(sentdict ,auxidx, tree, word_positions_in_tree): # 8% recall traded for 4% precision.
#if (not isverb(sentdict['pos'][auxidx+1])) or (sentdict['lemmas'][auxidx+1]=='not' and not isverb(sentdict['pos'][auxidx+2])):
return True
# if not auxccommandsverbthatcomesafter(sentdict ,auxidx, tree, word_positions_in_tree):
# return True
if verbfollowsaux(sentdict, auxidx):
return False
if isprep(sentdict['pos'][auxidx+1]) and sentdict['words'][auxidx] != 'done':
return True
return False
def auxccommandsverbthatcomesafter(sentdict, auxidx, tree, word_positions_in_tree):
for i in range(auxidx+1, len(sentdict['pos'])):
if isverb(sentdict['pos'][i]):
if nltktree.ccommands(tree, tree[word_positions_in_tree[auxidx-1]], tree[word_positions_in_tree[i-1]]):
return True
return False
def aux_ccommands_verb_that_comes_after(sentdict, aux, tree, word_positions_in_tree):
for i in range(aux.wordnum+1, len(sentdict)):
if is_verb(sentdict.pos[i]):
if nt.ccommands(tree, tree[word_positions_in_tree[aux.wordnum-1]], tree[word_positions_in_tree[i-1]]):
return True
return False
def ccommands(w1_idx, w2_idx, tree, word_positions):
try:
return nt.ccommands(tree, word_positions[w1_idx-1], word_positions[w2_idx-1])
except IndexError:
return False