Install Mecab and CaboCha on ubuntu16.04LTS so that it can be used from python3 series

Introduction

I found a library called cabocha that can perform dependency analysis, so I tried to play with it for the time being, but if I started touching it with a light feeling, I had a hard time even installing it ... For the time being, I would like to write that I could do this in my environment.

environment

• windows 10

• VirtualBox 5.1

• Ubuntu16.04 LTS

• python 3.5.2

• Ubuntu is installed in the VM on the windows machine

Main subject

Download MeCab and dictionary data

Execute the following command in the terminal

$ sudo apt-get -y install mecab libmecab-dev mecab-ipadic-utf8 mecab-jumandic-utf8

Binding of MeCab to python

• Added 2017/11/06

pip3 install mecab-python3

It seems better to do this just in case

Download CRF ++

Here Download the latest version from (.tar). Currently it was ver. 0.58. Unzip in a suitable directory. Move.

$ tar zxvf CRF++-0.58.tar.gz
$ cd CRF++-0.58

Fixed because there is a bug

$ vim node.cpp
#include <time.h>(Addition)

Save, close and install

$ ./configure
$ sudo make
$ sudo make install

Download CaboCha

Download from here. However, the latest version (cabocha-0.69.tar.bz2) failed. There is no choice but to download cabocha-0.60.tar.gz. Unzip in a suitable directory. Move.

$ tar xzvf cabocha-0.60.tar.gz
$ cd cabocha-0.60

Fixed because there is a bug

$ sudo vim /etc/ld.so.conf
include /usr/local/bin      (Postscript)

Run

sudo /sbin/ldconfig

Another fix

$ vim src/utils.cpp

utils.cpp

void Unlink(const char *filename) {
#if defined(_WIN32) && !defined(__CYGWIN__)
  ::DeleteFileA(filename);
#else
  //::unlink(filename);Delete
  ::remove(filename);   //Postscript
#endif
}

Save, close and install

$ ./configure --with-mecab-config=`which mecab-config` --with-charset=utf8
$ sudo make clean
$ sudo make
$ sudo make install
$ sudo /sbin/ldconfig

Try running cabocha

$ cabocha
$ (Favorite sentences)

Binding to python3 series

$ sudo apt-get install swig python3-dev
$ cd cabocha-0.60
$ swig -python -shadow -c++ swig/CaboCha.i
$ mv swig/CaboCha.py python/
$ mv swig/CaboCha_wrap.cxx python/

Rewritten to support python3 series

$ cd python
$ vim setup.py

setup.py

def cmd2(str):
    #return string.split (cmd1(str))Delete
    return cmd1(str).split()   #Postscript

Save, close and run

$ sudo python3 setup.py build_ext
$ sudo python3 setup.py install
$ sudo /sbin/ldconfig

• Note that if you are using pyenv, the installation location may be incorrect.

$ python3
$ import CaboCha

OK if no error occurs. This completes the installation. Please play appropriately.

in conclusion

Thank you for your hard work. It feels like I touched it lightly, and the accuracy is quite good, and it took some time to install it, but there seems to be a merit worth it. I would like to write a separate article about the results of playing around with it. Let's have a good natural language processing life!

reference

・ Http://taku910.github.io/mecab/ ・ Https://taku910.github.io/cabocha/ ・ Http://qiita.com/nezuq/items/f481f07fc0576b38e81d ・ Http://azwoo.hatenablog.com/entry/2015/10/01/234434 ・ Https://www.trifields.jp/install-cabocha-in-ubuntu-1038