An 'Rcpp' Interface for Eunjeon Project.

RmecabKo

The goal of RmecabKo is to parse Korean phrases with mecab-ko (Eunjeon project, and to provide helper functions to analyze Korean documents. RmecabKo provides R wrapper function of mecab-ko with Rcpp (in Mac OSX and Linux) or wrapper function of binary build of mecab-ko-msvc by system commands and file I/O (in Windows).

For instructions in Korean, refer to readme.rmd.

Installation

Mac OSX, Linux

First, install mecab-ko from the Bitbucket repository.

You can download a source of mecab-ko from Download page.

In Mac OSX terminal:

$ tar zxfv mecab-ko-XX.tar.gz
$ cd mecab-ko-XX
$ ./configure 
$ make
$ make check
$ sudo make install

In Linux:

$ tar zxfv mecab-ko-XX.tar.gz
$ cd mecab-ko-XX
$ ./configure 
$ make
$ make check
$ su
# make install

After the installation of mecab-ko, You can install RmecabKo from github with:

install.packages("RmecabKo")
# or, install.packages("devtools")
devtools::install_github("junhewk/RmecabKo")

You need to install mecab-ko-dic also.

(In Github version, install_dic function is added to support this functionality. You can install mecab-ko-dic with install_dic(). I'm working with custom dictionary function, for it mecab-ko-dic has to be installed by this function.)

Refer to Bitbucket page. The installation procedure is same as mecab-ko.

Windows

In Windows, install_mecab function is provided. You need to specify the installation path of the mecab-ko and meccab-ko-dic in the function parameter, mecabLocation.

install.packages("RmecabKo")
# install.packages("devtools")
devtools::install_github("junhewk/RmecabKo")
install_mecab("D:/Rlibs/mecab")

• 12/12/17. Encoding problem in Windows resolved (special thanks to Young Woo Kim).

Example

Basic usage of the provided functions is to put character vector in phrase parameter of pos(phrase) and nouns(phrase). Loop between phrases are operated in the C++ binary, thus you can analyze many phrases quickly.

pos("Hello. This is R wrapper of Korean morpheme analyzer mecab-ko.")

Output of the pos is list. Each element of the list contains classified morpheme and inferred part-of-speech (POS), separated by "/". The name of the element is the original phrase.

Output of the nouns is also list. Each element of the list contains extracted nouns. The name of the element is the original phrase.

tokenizer functions are added. You can use tokens_morph, tokens_words, tokens_nouns, and tokens_ngram. Please refer to the help page of each function.

More examples will be provided on Github wiki.

Author

Junhewk Kim ([email protected])

Thanks to and Contributor

• Eunjeon project: Fork Japanese morpheme analyzer mecab to Korean version
• Wonsup Yoon: VC++ binary build of mecab-ko-msvc, mecab-ko-dic-msvc.