def findNP(tree, sentNum, ngdata, mention_list, mentionID):
np_rels = ['su', 'app', 'body', 'sat', 'predc', 'obj1', 'cnj']
for mention_node in tree.findall(".//node[@cat='np']"):
len_ment = int(mention_node.attrib['end']) - int(
mention_node.attrib['begin'])
if mention_node.attrib['rel'] in np_rels and len_ment < 7:
name = 'np_' + mention_node.attrib['rel']
new_mention = make_mention(
mention_node.attrib['begin'],
mention_node.attrib['end'],
tree,
name,
sentNum,
next(mentionID),
ngdata,
)
subtrees = tree.findall('.//node[@begin="' +
str(new_mention.begin) + '"][@end="' +
str(new_mention.end) + '"]')
firstChild = subtrees[0].find('.//node[@word]')
if (('pron' in firstChild.attrib
and firstChild.attrib['pron'] == 'true')
or ('buiging' in firstChild.attrib
and firstChild.attrib['buiging'] == 'zonder')
or ('pos' in firstChild.attrib
and firstChild.attrib['pos'] == 'num')
or ('index' in firstChild.attrib)
or ('sc' in firstChild.attrib)):
pass
elif ',' in new_mention.tokenList:
handleComma(new_mention, tree, sentNum, ngdata, mention_list,
mentionID)
elif allWordsHaveAlpha(new_mention.tokenList):
add_mention(mention_list, new_mention)
def findPron(tree, sentNum, ngdata, mention_list, mentionID):
for mention_node in tree.findall(".//node[@pdtype='pron']") + tree.findall(
".//node[@frame='determiner(pron)']"):
new_mention = make_mention(
mention_node.attrib['begin'],
mention_node.attrib['end'],
tree,
'Pronoun',
sentNum,
next(mentionID),
ngdata,
)
add_mention(mention_list, new_mention)
def findObj(tree, sentNum, ngdata, mention_list, mentionID):
for mention_node in tree.findall(
".//node[@word][@ntype='soort'][@rel='obj1']"):
new_mention = make_mention(
mention_node.attrib['begin'],
mention_node.attrib['end'],
tree,
'noun',
sentNum,
next(mentionID),
ngdata,
)
add_mention(mention_list, new_mention)
def findSubj(tree, sentNum, ngdata, mention_list, mentionID):
for mention_node in tree.findall(".//node[@rel='su']"):
if 'cat' not in mention_node.attrib:
new_mention = make_mention(
mention_node.attrib['begin'],
mention_node.attrib['end'],
tree,
'su',
sentNum,
next(mentionID),
ngdata,
)
add_mention(mention_list, new_mention)
def findMWU(tree, sentNum, ngdata, mention_list, mentionID):
mwu_rels = ['obj1', 'su', 'cnj', 'hd'] # hd 14/65
for mention_node in tree.findall(".//node[@cat='mwu']"):
if mention_node.attrib['rel'] in mwu_rels:
name = 'mwu_' + mention_node.attrib['rel']
new_mention = make_mention(
mention_node.attrib['begin'],
mention_node.attrib['end'],
tree,
name,
sentNum,
next(mentionID),
ngdata,
)
add_mention(mention_list, new_mention)
def findMWU2(tree, sentNum, ngdata, mention_list, mentionID):
mwu_rels = ['obj1', 'su', 'cnj', 'hd'] # hd 14/65
for mention_node in tree.findall(".//node[@cat='mwu']"):
if mention_node.attrib['rel'] in mwu_rels:
prevDet = tree.find(".//node[@pos='det'][@end='" +
mention_node.attrib['begin'] + "']")
if prevDet is not None:
name = 'mwu_' + mention_node.attrib['rel']
new_mention = make_mention(
int(mention_node.attrib['begin']) - 1,
mention_node.attrib['end'],
tree,
name,
sentNum,
next(mentionID),
ngdata,
)
add_mention(mention_list, new_mention)
def findNP2(tree, sentNum, ngdata, mention_list, mentionID):
np_rels = ['obj1', 'su', 'app', 'cnj', 'body', 'sat', 'predc']
for mention_node in tree.findall(".//node[@cat='np']"):
len_ment = (int(mention_node.attrib['end']) -
int(mention_node.attrib['begin']))
if (mention_node.attrib['rel'] in np_rels
and len_ment > 4): # and len_ment < 10:
for die in tree.findall(
".//node[@word='die']"): # @word='die' werkt beter
if (int(die.attrib['begin']) > int(
mention_node.attrib['begin'])
and int(die.attrib['end']) < int(
mention_node.attrib['end'])):
new_mention = make_mention(mention_node.attrib['begin'],
die.attrib['begin'], tree,
'die_np', sentNum,
next(mentionID), ngdata)
if allWordsHaveAlpha(new_mention.tokenList):
add_mention(mention_list, new_mention)
def handleComma(new_mention, tree, sentNum, ngdata, mention_list, mentionID):
if new_mention.tokenList[1] == ',' and len(new_mention.tokenList) > 3:
add_mention(mention_list, new_mention)
elif (len(new_mention.tokenList) == 3 and new_mention.tokenList[1] == ','
and 'neclass' in new_mention.tokenAttribs[0]
and new_mention.tokenAttribs[0]['neclass'] == 'LOC'):
add_mention(mention_list, new_mention)
new_mention1 = make_mention(
new_mention.tokenList.index(',') + new_mention.begin + 1,
new_mention.end,
tree,
'np_comma',
sentNum,
next(mentionID),
ngdata,
)
add_mention(mention_list, new_mention)
else:
new_mention1 = make_mention(
new_mention.begin,
new_mention.begin + new_mention.tokenList.index(','),
tree,
'np_comma',
sentNum,
next(mentionID),
ngdata,
)
new_mention2 = make_mention(
new_mention.tokenList.index(',') + new_mention.begin + 1,
new_mention.end,
tree,
'np_comma',
sentNum,
next(mentionID),
ngdata,
)
add_mention(mention_list, new_mention1)
add_mention(mention_list, new_mention2)
def findName(tree, sentNum, ngdata, mention_list, mentionID):
for new_mention in stitch_names(tree.findall(".//node[@pos='name']"), tree,
sentNum, mentionID, ngdata):
add_mention(mention_list, new_mention)