def prepare_kiwi(train_file):
"""
input: train file i.e. corpora
output: kiwi model
"""
numThread = 4
kiwi = Kiwi(numThread)
reader = ReaderExam(train_file)
minCount = 5
maxWordLength = 6
minScore = 0.25
kiwi.extractWords(reader.read, minCount, maxWordLength, minScore)
kiwi.prepare()
return kiwi
self.input.close()
self.output.close()
kiwi = Kiwi()
kiwi.load_user_dictionary(r'./server_project/test/userDict.txt')
kiwi.add_user_word('iXVDR', 'NNP', 3.0)
kiwi.prepare()
# handle = IOHandler(r'./server_project/test/input.txt', r'./server_project/test/result.txt')
# kiwi.analyze(handle.read, handle.write)
result = kiwi.analyze('강남에서 먹었던 오늘의 스파게티는 맛있었다.', 1)
for i in result:
print(i)
class ReaderExam:
def __init__(self, filePath):
self.file = open(filePath, encoding='UTF8')
def read(self, id):
if id == 0: self.file.seek(0)
return self.file.readline()
reader = ReaderExam(r'./server_project/test/input.txt')
print(kiwi.extractWords(reader.read, 1, 10, 0.25))
#kiwi.extract_add_words(reader.read, min_cnt = 1, max_word_len = 10, min_score = 0.25, pos_score = -3)
