mecab-ipadic-NEologd : Neologism dictionary for MeCab

For Japanese

README.ja.md is written in Japanese.

https://github.com/neologd/mecab-ipadic-neologd/blob/master/README.ja.md

Documentation

You can find more detailed documentation and examples in the following wiki.

https://github.com/neologd/mecab-ipadic-neologd/wiki/Home

Overview

mecab-ipadic-NEologd is customized system dictionary for MeCab.

This dictionary includes many neologisms (new word), which are extracted from many language resources on the Web.

When you analyze the Web documents, it's better to use this system dictionary and default one (ipadic) together.

Pros and Cons

Pros

Recorded about 2.525 million pairs(including duplicate entries) of surface/furigana(kana indicating the pronunciation of kanji) of the words such as the named entity that can not be tokenized correctly using default system dictionary of MeCab.
Update process of this dictionary will automatically run on development server.
- I'm planning to renew this dictionary at least updating twice weekly
  - Every Monday and Thursday
When renewing by utilizing the language resources on Web, a new named entity can be recorded.
- The resources are being utilized at present are as follows.
  - Dump data of hatena keyword
  - Japanese postal code number data download (ken_all.lzh)
  - The name-of-the-station list of whole country of Japan
  - The entry data of the person names (last name / first name)
  - The entry data of Unicode emojis (version 8.0)
  - The entry data of Kaomoji strings
  - The entry data of adverbs
  - The entry data of adjectives
  - The entry data of adjective verbs
  - The entry data of interjections
  - The entry data of orthographic variant of general nouns
  - The entry data of orthographic variant of nouns used as verb form
A lot of documents, which crawled from Web
- I'm planning to record the words such as the named entity, which will be extracted from other new language resource.

Cons

Classification of the named entity is insufficient
- Ex. Some person names and a product name are classified into the same named entity category.
Not named entity word is recorded as named entity too.
Since the manual checking to all the named entities can't conduct, it may have made a mistake in matching of surface of the named entity and furigana of the named entity.
Unless the language resources on Web are updated, a new named entity won't add to the dictionary.
Corresponding to only UTF-8
- You should install the UTF-8 version of MeCab

Getting started

Memory requirements

Required: 1.5GB of RAM
Recommend: 5GB of RAM
- Current maximum binary size is 900MB

Dependencies

We build mecab-ipadic-NEologd using the source code of mecab-ipadic at installing phase.

You should install following libraries using apt or yum, homebrew, source-code.

C++ Compiler
- Operation on GCC-4.4.7 and Apple LLVM version 6.0 are confirmed
iconv (libiconv)
- Use to convert the character code of IPADIC
mecab
- We use bin/mecab and bin/mecab-config
mecab-ipadic
- One of a dictionary of MeCab
  - Use to test at the time of installation
  - If you install it using source code, you should select the UTF-8 as a character code
./configure --with-charset=utf8; make; sudo make install
xz
- Use to decompress a xz package of a seed of mecab-ipadic-NEologd

Please install at any time other lack library.

Examples

On CentOS

$ sudo rpm -ivh http://packages.groonga.org/centos/groonga-release-1.1.0-1.noarch.rpm

$ sudo yum install mecab mecab-devel mecab-ipadic git make curl xz
On Fedora

$ sudo yum install mecab mecab-devel mecab-ipadic git make curl xz
On Ubuntu

$ sudo aptitude install mecab libmecab-dev mecab-ipadic-utf8 git make curl xz-utils
On MacOSX

$ brew install mecab mecab-ipadic git curl xz

Preparation of installing

A seed data of the dictionary will distribute via GitHub repository.

In first time, you should execute the following command to 'git clone'.

$ git clone --depth 1 https://github.com/neologd/mecab-ipadic-neologd.git

$ git clone --depth 1 [email protected]:neologd/mecab-ipadic-neologd.git

If you need all log of mecab-ipadic-neologd.git, you should clone the repository without '--depth 1'

How to install/update mecab-ipadic-NEologd

Step.1

Move to a directory of the repository which was cloned in the above preparation.

$ cd mecab-ipadic-neologd

Step.2

You can install or can update(overwritten) the recent mecab-ipadic-NEologd by following command.

$ ./bin/install-mecab-ipadic-neologd -n

You can check the directory path to install it.

$ echo `mecab-config --dicdir`"/mecab-ipadic-neologd"

If you installed some mecab-config, you should set the path of the mecab-config which you want to use.

If you use following command, you can check useful command line option.

$ ./bin/install-mecab-ipadic-neologd -h

How to use mecab-ipadic-NEologd

When you want to use mecab-ipadic-NEologd, you should set the path of custom system dictionay(*/lib/mecab/dic/mecab-ipadic-neologd/) as -d option of MeCab.

Example (on CentOS)

$ mecab -d /usr/local/lib/mecab/dic/mecab-ipadic-neologd/

Example of output of MeCab

When you use mecab-ipadic-NEologd

$echo "10日放送の「中居正広のミになる図書館」（テレビ朝日系）で、SMAPの中居正広が、篠原信一の過去の勘違いを明かす一幕があった。" | mecab -d /usr/local/lib/mecab/dic/mecab-ipadic-neologd
10日    名詞,固有名詞,一般,*,*,*,10日,トオカ,トオカ
放送    名詞,サ変接続,*,*,*,*,放送,ホウソウ,ホーソー
の      助詞,連体化,*,*,*,*,の,ノ,ノ
「      記号,括弧開,*,*,*,*,「,「,「
中居正広のミになる図書館        名詞,固有名詞,一般,*,*,*,中居正広のミになる図書館,ナカイマサヒロノミニナルトショカン,ナカイマサヒロノミニナルトショカン
」      記号,括弧閉,*,*,*,*,」,」,」
（      記号,括弧開,*,*,*,*,（,（,（
テレビ朝日      名詞,固有名詞,組織,*,*,*,テレビ朝日,テレビアサヒ,テレビアサヒ
系      名詞,接尾,一般,*,*,*,系,ケイ,ケイ
）      記号,括弧閉,*,*,*,*,）,）,）
で      助詞,格助詞,一般,*,*,*,で,デ,デ
、      記号,読点,*,*,*,*,、,、,、
SMAP    名詞,固有名詞,一般,*,*,*,SMAP,スマップ,スマップ
の      助詞,連体化,*,*,*,*,の,ノ,ノ
中居正広        名詞,固有名詞,人名,*,*,*,中居正広,ナカイマサヒロ,ナカイマサヒロ
が      助詞,格助詞,一般,*,*,*,が,ガ,ガ
、      記号,読点,*,*,*,*,、,、,、
篠原信一        名詞,固有名詞,人名,*,*,*,篠原信一,シノハラシンイチ,シノハラシンイチ
の      助詞,連体化,*,*,*,*,の,ノ,ノ
過去    名詞,副詞可能,*,*,*,*,過去,カコ,カコ
の      助詞,連体化,*,*,*,*,の,ノ,ノ
勘違い  名詞,サ変接続,*,*,*,*,勘違い,カンチガイ,カンチガイ
を      助詞,格助詞,一般,*,*,*,を,ヲ,ヲ
明かす  動詞,自立,*,*,五段・サ行,基本形,明かす,アカス,アカス
一幕    名詞,一般,*,*,*,*,一幕,ヒトマク,ヒトマク
が      助詞,格助詞,一般,*,*,*,が,ガ,ガ
あっ    動詞,自立,*,*,五段・ラ行,連用タ接続,ある,アッ,アッ
た      助動詞,*,*,*,特殊・タ,基本形,た,タ,タ
。      記号,句点,*,*,*,*,。,。,。
EOS

What's the point of the above result

MeCab can parse the words which are recorded in mecab-ipadic-NEologd correctly.
- "中居正広のミになる図書館(Masahiro Nakai's library to improve an ability)" is a new word.
  - This word could parse correctly because of updating of the language resources on Web.
Almost all of the entry of mecab-ipadic-NEologd has the value of furigana field.

When you use default system dictionary

$echo "10日放送の「中居正広のミになる図書館」（テレビ朝日系）で、SMAPの中居正広が、篠原信一の過去の勘違いを明かす一幕があった。" | mecab
10      名詞,数,*,*,*,*,*
日      名詞,接尾,助数詞,*,*,*,日,ニチ,ニチ
放送    名詞,サ変接続,*,*,*,*,放送,ホウソウ,ホーソー
の      助詞,連体化,*,*,*,*,の,ノ,ノ
「      記号,括弧開,*,*,*,*,「,「,「
中居    名詞,固有名詞,人名,姓,*,*,中居,ナカイ,ナカイ
正広    名詞,固有名詞,人名,名,*,*,正広,マサヒロ,マサヒロ
の      助詞,連体化,*,*,*,*,の,ノ,ノ
ミ      名詞,一般,*,*,*,*,ミ,ミ,ミ
に      助詞,格助詞,一般,*,*,*,に,ニ,ニ
なる    動詞,自立,*,*,五段・ラ行,基本形,なる,ナル,ナル
図書館  名詞,一般,*,*,*,*,図書館,トショカン,トショカン
」      記号,括弧閉,*,*,*,*,」,」,」
（      記号,括弧開,*,*,*,*,（,（,（
テレビ朝日      名詞,固有名詞,組織,*,*,*,テレビ朝日,テレビアサヒ,テレビアサヒ
系      名詞,接尾,一般,*,*,*,系,ケイ,ケイ
）      記号,括弧閉,*,*,*,*,）,）,）
で      助詞,格助詞,一般,*,*,*,で,デ,デ
、      記号,読点,*,*,*,*,、,、,、
SMAP    名詞,固有名詞,組織,*,*,*,*
の      助詞,連体化,*,*,*,*,の,ノ,ノ
中居    名詞,固有名詞,人名,姓,*,*,中居,ナカイ,ナカイ
正広    名詞,固有名詞,人名,名,*,*,正広,マサヒロ,マサヒロ
が      助詞,格助詞,一般,*,*,*,が,ガ,ガ
、      記号,読点,*,*,*,*,、,、,、
篠原    名詞,固有名詞,人名,姓,*,*,篠原,シノハラ,シノハラ
信一    名詞,固有名詞,人名,名,*,*,信一,シンイチ,シンイチ
の      助詞,連体化,*,*,*,*,の,ノ,ノ
過去    名詞,副詞可能,*,*,*,*,過去,カコ,カコ
の      助詞,連体化,*,*,*,*,の,ノ,ノ
勘違い  名詞,サ変接続,*,*,*,*,勘違い,カンチガイ,カンチガイ
を      助詞,格助詞,一般,*,*,*,を,ヲ,ヲ
明かす  動詞,自立,*,*,五段・サ行,基本形,明かす,アカス,アカス
一幕    名詞,一般,*,*,*,*,一幕,ヒトマク,ヒトマク
が      助詞,格助詞,一般,*,*,*,が,ガ,ガ
あっ    動詞,自立,*,*,五段・ラ行,連用タ接続,ある,アッ,アッ
た      助動詞,*,*,*,特殊・タ,基本形,た,タ,タ
。      記号,句点,*,*,*,*,。,。,。
EOS

To evaluate or to reproduce a research result

We release the tag for mainly research purpose.

https://github.com/neologd/mecab-ipadic-neologd/releases/

It is very useful for the following applications.

Experiments for evaluation of a research result
Reproducibility of a experimental result of others
Creation of the processing results of morphological analysis that doesn't update forever

If you are the beginner of NLP, I recommend that you use the latest version of master branch.

Bibtex

Please use the following bibtex, when you refer mecab-ipadic-NEologd from your papers.

@misc{sato2015mecabipadicneologd,
    title  = {Neologism dictionary based on the language resources on the Web for Mecab},
    author = {Toshinori, Sato},
    url    = {https://github.com/neologd/mecab-ipadic-neologd},
    year   = {2015}
}

Star please !!

Please star this github repository if mecab-ipadic-NEologd is very useful to your project ;)

Copyrights

We select the 'Apache License, Version 2.0'. Please check following link.

https://github.com/neologd/mecab-ipadic-neologd/blob/master/COPYING

romaryd / mecab-ipadic-neologd Goto Github PK

mecab-ipadic-neologd's Introduction