def main() -> None:
for chunks in load_chunk_cabocha():
for chunk in chunks:
if chunk.dst == -1:
continue
# 名詞と動詞を含む文節のフラグ
exist_noun = False
exist_verb = False
# 現在の文節内に名詞が存在するかの判定
for m in chunk.morphs:
if m.pos == "名詞":
exist_noun = True
# 現在の文節の掛かり先が動詞であるかの判定
for m in chunks[chunk.dst].morphs:
if m.pos == "動詞":
exist_verb = True
# 両方のフラグが True の場合に出力する
if exist_noun and exist_verb:
src = chunk.no_symbols()
dst = chunks[chunk.dst].no_symbols()
print(f"{src}\t{dst}")
def get_case_pattern_of_verb() -> None:
for sentence in load_chunk_cabocha():
case_pattern = {} # type: Dict[int, Dict[str, List[str]]]
# 例
# {1: {'生れる': ['で']}, 4: {'つく': ['か', 'が']}}
# {5: {'泣く': ['で']}, 7: {'する': ['て', 'だけ', 'は']}}
for chunk in sentence:
if chunk.dst == -1:
continue
# 助詞の探索
pp = [m.surface for m in chunk.morphs if m.pos == "助詞"]
# 述語の探索 (助詞の掛かり先から探索する)
verbs = [
m.base for m in sentence[chunk.dst].morphs if m.pos == "動詞"
]
if verbs == [] or pp == []:
continue
if chunk.dst not in case_pattern:
# 最左の動詞をキーとした助詞のリストを、文節の係り先をキーとして追加する
case_pattern[chunk.dst] = {verbs[0]: pp}
else:
# 既存の場合、助詞のリストを追加する
case_pattern[chunk.dst][verbs[0]].extend(pp)
for dic in case_pattern.values():
for verb, pp in dic.items():
print(f"{verb}\t{' '.join(sorted(pp))}")
def get_case_frame_of_verb() -> None:
for sentence in load_chunk_cabocha():
case_pattern = [] # type: List[CasePattern]
# 助詞を含む文節を探索していく
for chunk in sentence:
if chunk.dst == -1:
continue
# 助詞の探索
pp = [m.surface for m in chunk.morphs if m.pos == "助詞"]
# 述語の探索 (助詞の係り先から探索する)
verbs = [
m.base for m in sentence[chunk.dst].morphs if m.pos == "動詞"
]
if verbs == [] or pp == []:
continue
# 文節の係り先と一致する case.id のリスト
if [case.id for case in case_pattern
if case.id == chunk.dst] == []:
# case.id に文節の係り先を保存する
case_pattern.append(
CasePattern(chunk.dst, verbs[0], pp, chunk.no_symbols()))
else:
# 同じ述語へ係る格を追加する
for case in case_pattern:
if case.id == chunk.dst:
case.pp.extend(pp)
case.chunk.append(chunk.no_symbols())
break
def main() -> None:
for chunks in load_chunk_cabocha():
for chunk in chunks:
if chunk.dst == -1:
continue
# Chunk.no_symbols() で記号を取り除いた形式にする
src = chunk.no_symbols()
dst = chunks[chunk.dst].no_symbols()
# src, dst が空でなければ出力
if src != "" and dst != "":
print(f"{src} {dst}")
def get_path_from_noun_to_root() -> None:
for chunks in load_chunk_cabocha():
for chunk in chunks:
if chunk.dst == -1:
continue
# 文節に名詞が存在しない
if not any([m.pos == "名詞" for m in chunk.morphs]):
continue
dst = chunk.dst
phrases = [chunk.no_symbols()]
# 係り先 (Chunk.dst) が -1 になるまで名詞からパスを探索する
while chunks[dst].dst != -1:
phrases.append(chunks[dst].no_symbols())
dst = chunks[dst].dst
print(" -> ".join(phrases))
def get_dependency_path_between_nouns() -> None:
for chunks in load_chunk_cabocha():
phrase_path = defaultdict(list) # type: PD
for i, chunk in enumerate(chunks):
if not any(m.pos == "名詞" for m in chunk.morphs):
continue
if chunk.dst == -1:
phrase_path[i] = []
continue
current = chunk
while current.dst != -1:
phrase_path[i].append(current.dst)
current = chunks[current.dst]
for i, j in combinations(phrase_path.keys(), 2):
print(join_phrase(i, j, phrase_path, chunks))
def mining_light_verb_syntax():
for chunks in load_chunk_cabocha():
for i, chunk in enumerate(chunks):
if chunk.srcs == []:
continue
# 動詞の探索
verbs = [morph.base for morph in chunk.morphs if morph.pos == "動詞"]
if verbs == []:
continue
# 助詞を含む文節のリスト
phrases_containing_particle = []
for src in chunk.srcs:
# 参照先の文節に助詞が一つでも含まれていれば、参照先の文節を追加する
if any(morph.pos == "助詞" for morph in chunks[src].morphs):
phrases_containing_particle.append(chunks[src])
if phrases_containing_particle == []:
continue
# 「サ変接続名詞+を」となる文節を探し助詞を含む文節のリストから除去する
light_verb = ""
for phrase in phrases_containing_particle:
for i in range(len(phrase.morphs) - 1):
if (phrase.morphs[i].pos1 == "サ変接続"
and phrase.morphs[i + 1].surface == "を"):
light_verb = f"{phrase.morphs[i].surface}を{verbs[0]}"
phrases_containing_particle.remove(phrase)
break
if light_verb == "":
continue
# 助詞と文節のリストの作成
particles_and_phrases = []
for phrase in phrases_containing_particle:
for morph in phrase.morphs:
if morph.pos == "助詞":
particles_and_phrases.append(
[morph.surface, phrase.no_symbols()])
break
particles = [pp[0] for pp in sorted(particles_and_phrases)]
phrases = [pp[1] for pp in sorted(particles_and_phrases)]
print(f"{light_verb}\t{' '.join(particles)}\t{' '.join(phrases)}")
def main(n=10) -> None:
chunks_list = load_chunk_cabocha()
visualize_dependency_tree(chunks_list[n])