def get_all_sememes(self):
"""
Get the complete sememe list in HowNet
:return: (List) a list of sememes
"""
if hasattr(self, "sememe_all"):
return self.sememe_all
else:
package_directory = os.path.dirname(os.path.abspath(__file__))
f = get_resource("sememe_all.txt", 'r')
buf = f.readlines()[0]
self.sememe_all = buf.strip().split()
return self.sememe_all
def _load_taxonomy(self):
package_directory = os.path.dirname(os.path.abspath(__file__))
f = get_resource("sememe_triples_taxonomy.txt", "r")
self.sememe_taxonomy = {}
self.sememe_dict = {}
for line in f.readlines():
line = line.strip().split(" ")
for u in line[0].split("|"):
for v in line[2].split("|"):
self.sememe_taxonomy[(u, v)] = line[1]
if not (u, line[1]) in self.sememe_dict:
self.sememe_dict[(u, line[1])] = []
self.sememe_dict[(u, line[1])].append(line[2])
def __init__(self, use_sim=False):
'''
Initialize HowNetDict
:param use_sim: "lazy" option for loading similarity computation file.
'''
try:
package_directory = os.path.dirname(os.path.abspath(__file__))
data_dir = os.path.join(package_directory, "HowNet_dict_complete")
self.en_map = dict()
self.zh_map = dict()
self.ids = dict()
# load dict complete
with get_resource(data_dir, 'rb') as origin_dict:
word_dict = pickle.load(origin_dict)
# self.max_count = len(word_dict) + 10
for key in word_dict:
now_dict = word_dict[key]
en_word = now_dict["en_word"].strip()
zh_word = now_dict["ch_word"].strip()
if en_word not in self.en_map:
self.en_map[en_word] = list()
self.en_map[en_word].append(now_dict)
if zh_word not in self.zh_map:
self.zh_map[zh_word] = list()
self.zh_map[zh_word].append(now_dict)
if now_dict["No"] not in self.ids:
self.ids[now_dict["No"]] = list()
self.ids[now_dict['No']].append(now_dict)
# self.ids[now_dict["No"]] = now_dict
# else:
# now_dict['No'] = str(self.max_count)
# self.ids[now_dict["No"]] = now_dict
# self.max_count += 1
if use_sim:
if not self.initialize_sememe_similarity_calculation():
self.hownet = None
self.sememe_root = None
self.sememe_sim_table = None
except FileNotFoundError as e:
print(e)