1 minute read

MeCab

MeCab is an open source Japanese Part-Of-Speech tagging engine developed and maintained by Dr. Taku Kudo.

Requirements

  1. g++ 3.4.3 or later
  2. iconv

Install from Source

Use the tar command to extract mecab-0.996.tar.gz.

$ tar zxvf mecab-0.996.tar.gz

cd into mecab-0.966 and use the following commands to install MeCab.

$ cd mecab-0.996
$ ./configure --with-charset=UTF8
$ make
$ make check
$ sudo make install

Use again the tar command to extract mecab-ipadic-2.7.0-20070801.tar.gz, the Japanese dictionary for MeCab. Then cd into mecab-ipadic-2.7.0-20070801 and install the ipadic dictionary as follows.

$ tar zxvf mecab-ipadic-2.7.0-20070801.tar.gz
$ cd mecab-ipadic-2.7.0-20070801
$ ./configure --with-charset=UTF8
$ make
$ sudo make install

If ./configure --with-charset=UTF8 gives an error, try the following command

sudo ldconfig

to update the shared library system.

Refer to the following in case you want to use MeCab in Python.

$ tar zxvf mecab-python-0.996.tar.gz
$ cd mecab-python-0.996
$ python setup.py build
$ python setup.py install

CaboCha

CaboCha is an open source Japanese depedency parser also developed and maintained by Dr. Kudo.

Requirements

  1. MeCab 0.993 or later
  2. One of the follwing dictionary for MeCab: mecab-ipadic, mecab-jamandic and unidic

Install from Source

Run the commands below to install the conditional random fields toolkit CRF++ used as dependency for CaboCha.

$ tar zxvf CRF++-0.58.tar.gz
$ cd CRF++-0.58
$ ./configure
$ make
$ sudo make install

Now is the time to install CaboCha. Use the tar command to extract cabocha-0.67.tar.bz2

tar jxvf cabocha-0.67.tar.bz2

Move to cabocha-0.67 and use the following commands to install CaboCha.

$ tar jxvf cabocha-0.67.tar.gz
$ cd cabocha-0.67
$ ./configure --with-charset=UTF8
$ make
$ make check
$ sudo make install

Furthermore

$ cd python
$ python setup.py build
$ python setup.py install

if you want to use CaboCha from Python. After installing the package, you may be required to execute the ldconfig command again in case you fail to import the CaboCha module from the Python intepreter.