python-mecab-ko 1.3.0

A python binding for mecab-ko

python-mecab-ko

A python binding for mecab-ko

Installation

Using pip:

$ pip install python-mecab-ko

This library currently provides binary wheels on Linux, macOS and Windows platforms. If you are using other platforms or would like to build from the source, you need the following prerequisites:

• C++14 compatible compiler
• Python header files
• mecab-ko
  • This repository provides a convenience script at scripts/install_mecab_ko.py to install mecab-ko

Usage

import mecab

mecab = mecab.MeCab()

mecab.morphs('영등포구청역에 있는 맛집 좀 알려주세요.')
# ['영등포구청역', '에', '있', '는', '맛집', '좀', '알려', '주', '세요', '.']

mecab.nouns('우리나라에는 무릎 치료를 잘하는 정형외과가 없는가!')
# ['우리', '나라', '무릎', '치료', '정형외과']

mecab.pos('자연주의 쇼핑몰은 어떤 곳인가?')
# [('자연주의', 'NNG'), ('쇼핑몰', 'NNG'), ('은', 'JX'), ('어떤', 'MM'), ('곳', 'NNG'), ('인가', 'VCP+EF'), ('?', 'SF')]

mecab.parse('즐거운 하루 보내세요!')
# [
#     Morpheme(span=Span(start=0, end=3), surface="즐거운",
#         feature=Feature(
#             pos="VA+ETM", semantic=None, has_jongseong=True, reading="즐거운",
#             type="Inflect", start_pos="VA", end_pos="ETM", exprssion="즐겁/VA/*+ᆫ/ETM/*",
#         ),
#     ),
#     Morpheme(span=Span(start=4, end=6), surface="하루",
#         feature=Feature(
#             pos="NNG", semantic=None, has_jongseong=False, reading="하루",
#             type=None, start_pos=None, end_pos=None, exprssion=None,
#         ),
#     ),
#     Morpheme(
#         span=Span(start=7, end=9), surface="보내",
#         feature=Feature(
#             pos="VV", semantic=None, has_jongseong=False, reading="보내",
#             type=None, start_pos=None, end_pos=None, exprssion=None,
#         ),
#     ),
#     Morpheme(
#         span=Span(start=9, end=11), surface="세요",
#         feature=Feature(
#             pos="EP+EF", semantic=None, has_jongseong=False, reading="세요",
#             type="Inflect", start_pos="EP", end_pos="EF", exprssion="시/EP/*+어요/EF/*",
#         ),
#     ),
#     Morpheme(
#         span=Span(start=11, end=12), surface="!",
#         feature=Feature(
#             pos="SF", semantic=None, has_jongseong=None, reading=None,
#             type=None, start_pos=None, end_pos=None, exprssion=None,
#         ),
#     ),
# ]

CLI

The library also provides a proxy for several command-line interfaces provided by mecab. These interfaces can be used to add vocabulary to a user dictonary.

• python3 -m mecab
• python3 -m mecab dict-index
• python3 -m mecab dict-gen
• python3 -m mecab cost-train

User Dictionary

You can build a user dictionary using the above command-line interface:

$ python3 -m mecab dict-index \
    --userdic user.dic \
    user.csv

NOTE

• The CSV file must be in the following format:
  • <surface>,<left_context_id>,<right_context_id>,<cost>,<pos>,<semantic>,<has_jongseong>,<reading>,<type>,<start_pos>,<end_pos>,<expression>
• Example:
  • 트위치,,,,NNP,*,F,트위치,*,*,*,*
  • 플랫폼,,,,NNG,*,T,플랫폼,*,*,*,*

Then, you can add the built user dictionary as follows:

from mecab import MeCab

mecab = MeCab(user_dictionary_path="user.dic")

# If there are multiple dictionaries:
mecab = MeCab(user_dictionary_path=["nnp.dic", "nng.dic"])

Fore more detailed information, please refer to examples/user_dictionary.py.

Acknowledgments

• APIs are inspired by KoNLPy