Japanese dependency analyzer. Used for natural language processing.
As you can see on the Official Site, basically you should download the source and enter it with the following command. You also need to install CRF ++ and Mecab in advance.
./configure
make
make check
sudo make install
If you want to use it with python, you can also do the following in the python folder.
python setup.py install
The official download link jumps to the public folder of Google Drive, so I didn't know for a moment what to do with the URL specified for wget when putting it in CentOS. For the time being, I was able to download it below. (Only CRF ++ was authenticated to googlecode ...)
| - | url |
|---|---|
| Mecab | http://cabocha.googlecode.com/files/cabocha-0.996.tar.bz2 |
| CRF++ | https://googledrive.com/host/0B4y35FiV1wh7fngteFhHQUN2Y1B5eUJBNHZUemJYQV9VWlBUb3JlX0xBdWVZTWtSbVBneU0/CRF++-0.58.tar.gz |
| Cabocha | http://cabocha.googlecode.com/files/cabocha-0.996.tar.bz2 |
iconv conversion failed. skip this entry.... I think that the following warning (?) Will appear in a row when you make it.
iconv conversion failed. skip this entry....
It seems to be a problem with encode. In ./configure, the options are:
./configure --with-charset=utf8 --enable-utf8-only
I get this kind of error when making
error while loading shared libraries: libcrfpp.so.0: cannot open shared object file: No such file or directory
It depends on the environment, but it's okay if you add the dependency information.
echo "/usr/local/lib" >> /etc/ld.so.conf.d/lib.conf
ldconfig
I think string.split (cmd1 (str)) will somehow get angry.
This error occurs because the cabocha code does not support python3.
A kind person wrote the patch file, so please modify the source accordingly.
For the time being, like this. I might add it if something happens again.
Recommended Posts