def find_ngram(self, wordclass):
"""
Return the ngram corresponding to the wordclass specified;
whether this is a Penn wordclass or a native wordclass
('NOUN', 'VERB', etc.).
"""
if wordclass in self.ngrams_by_penn():
return self.ngrams_by_penn()[wordclass]
elif wordclass_base(wordclass) in self.ngrams_by_penn():
return self.ngrams_by_penn()[wordclass_base(wordclass)]
else:
for ngram in self.ngrams_list:
if ngram.wordclass == wordclass:
return ngram
return None
def subcategory_ratio(self, pos):
baseclass = wordclass_base(pos)
base_ratio = self.ratio(baseclass)
if pos in self.pos_to_lemma_ratios:
return base_ratio / self.pos_to_lemma_ratios[pos][1]
else:
return base_ratio
def base_ratios(self):
try:
return self._base_ratios
except AttributeError:
self._base_ratios = defaultdict(lambda: 0)
for pos, value in self.ratios().items():
wc = wordclass_base(pos)
self._base_ratios[wc] += value
return self._base_ratios
def sum_subcategories(self, parts):
if parts:
baseclass = wordclass_base(parts[0])
base_ratio = self.ratio(baseclass)
total = 0
for pos in parts:
if pos in self.pos_to_lemma_ratios:
total += base_ratio / self.pos_to_lemma_ratios[pos][1]
else:
total += base_ratio
return min((total, base_ratio))
else:
return 0
def add_np(self):
"""
Check if a dummy "NP" wordclass needs to be added
"""
if (not 'NP' in self.full_set_of_wordclasses() and
self.wordform.lower() != self.wordform and
is_proper_name(self.wordform)):
group_type = wordclass_group('NP')
base_class = wordclass_base('NP')
if not group_type in self.model():
self.model()[group_type] = Group('core')
self.model()[group_type].add_null(base_class)
self.model()[group_type].model()[base_class].add_null('NP')
self.np_added = True
def base_set_of_wordclasses(self):
return set([wordclass_base(wc)
for wc in self.full_set_of_wordclasses()])
def __init__(self, wordclass):
self.wordclass = wordclass
self.baseclass = wordclass_base(wordclass)
self.items = list()
self.ratio = 1.0 # default placeholder
def add(self, lex_item):
base = wordclass_base(lex_item.wordclass)
if not base in self.model():
self.model()[base] = Base(base)
self.model()[base].add(lex_item)
