12.9.9 MeCab Full-Text Parser Plugin

The built-in MySQL full-text parser uses the white space between words as a delimiter to determine where words begin and end, which is a limitation when working with ideographic languages that do not use word delimiters. To address this limitation for Japanese, MySQL provides a MeCab full-text parser plugin. The MeCab full-text parser plugin is supported for use with InnoDB and MyISAM.

The MeCab full-text parser plugin is a full-text parser plugin for Japanese that tokenizes a sequence of text into meaningful words. For example, MeCab tokenizes “データベース管理” (“Database Management”) into “データベース” (“Database”) and “管理” (“Management”). By comparison, the ngram full-text parser tokenizes text into a contiguous sequence of n characters, where n represents a number between 1 and 10.

In addition to tokenizing text into meaningful words, MeCab indexes are typically smaller than ngram indexes, and MeCab full-text searches are generally faster. One drawback is that it may take longer for the MeCab full-text parser to tokenize documents, compared to the ngram full-text parser.

The full-text search syntax described in Section 12.9, “Full-Text Search Functions” applies to the MeCab parser plugin. Differences in parsing behavior are described in this section. Full-text related configuration options are also applicable.

For additional information about the MeCab parser, refer to the MeCab: Yet Another Part-of-Speech and Morphological Analyzer project on Github.

Installing the MeCab Parser Plugin

The MeCab parser plugin requires mecab and mecab-ipadic.

On supported Fedora, Debian and Ubuntu platforms (except Ubuntu 12.04 where the system mecab version is too old), MySQL dynamically links to the system mecab installation if it is installed to the default location. On other supported Unix-like platforms, libmecab.so is statically linked in libpluginmecab.so, which is located in the MySQL plugin directory. mecab-ipadic is included in MySQL binaries and is located in MYSQL_HOME\lib\mecab.

You can install mecab and mecab-ipadic using a native package management utility (on Fedora, Debian, and Ubuntu), or you can build mecab and mecab-ipadic from source. For information about installing mecab and mecab-ipadic using a native package management utility, see Installing MeCab From a Binary Distribution (Optional). If you want to build mecab and mecab-ipadic from source, see Building MeCab From Source (Optional).

On Windows, libmecab.dll is found in the MySQL bin directory. mecab-ipadic is located in MYSQL_HOME/lib/mecab.

To install and configure the MeCab parser plugin, perform the following steps:

Contents Haut

Creating a FULLTEXT Index that uses the MeCab Parser

To create a FULLTEXT index that uses the mecab parser, specify WITH PARSER ngram with CREATE TABLE, ALTER TABLE, or CREATE INDEX.

This example demonstrates creating a table with a mecab FULLTEXT index, inserting sample data, and viewing tokenized data in the INFORMATION_SCHEMA.INNODB_FT_INDEX_CACHE table:

mysql> USE test;

mysql> CREATE TABLE articles (
      id INT UNSIGNED AUTO_INCREMENT NOT NULL PRIMARY KEY,
      title VARCHAR(200),
      body TEXT,
      FULLTEXT (title,body) WITH PARSER mecab
    ) ENGINE=InnoDB CHARACTER SET utf8;

mysql> SET NAMES utf8;
    
mysql> INSERT INTO articles (title,body) VALUES
    ('データベース管理','このチュートリアルでは、私はどのようにデータベースを管理する方法を紹介します'),
    ('データベースアプリケーション開発','データベースアプリケーションを開発することを学ぶ');

mysql> SET GLOBAL innodb_ft_aux_table="test/articles";

mysql> SELECT * FROM INFORMATION_SCHEMA.INNODB_FT_INDEX_CACHE ORDER BY doc_id, position;

To add a FULLTEXT index to an existing table, you can use ALTER TABLE or CREATE INDEX. For example:

Contents Haut

MeCab Parser Space Handling

The MeCab parser uses spaces as separators in query strings. For example, the MeCab parser tokenizes データベース管理 as データベース and 管理.

Contents Haut

MeCab Parser Stopword Handling

By default, the MeCab parser uses the default stopword list, which contains a short list of English stopwords. For a stopword list applicable to Japanese, you must create your own. For information about creating stopword lists, see Section 12.9.4, “Full-Text Stopwords”.

Contents Haut

MeCab Parser Term Search

For natural language mode search, the search term is converted to a union of tokens. For example, データベース管理 is converted to データベース 管理.

SELECT COUNT(*) FROM articles WHERE MATCH(title,body) AGAINST('データベース管理' IN NATURAL LANGUAGE MODE);

For boolean mode search, the search term is converted to a search phrase. For example, データベース管理 is converted to データベース 管理.

SELECT COUNT(*) FROM articles WHERE MATCH(title,body) AGAINST('データベース管理' IN BOOLEAN MODE);

Contents Haut

MeCab Parser Wildcard Search

Wildcard search terms are not tokenized. A search on データベース管理* is performed on the prefix, データベース管理.

SELECT COUNT(*) FROM articles WHERE MATCH(title,body) AGAINST('データベース*' IN BOOLEAN MODE);

Contents Haut

MeCab Parser Phrase Search

Phrases are tokenized. For example, データベース管理 is tokenized as データベース 管理.

SELECT COUNT(*) FROM articles WHERE MATCH(title,body) AGAINST('"データベース管理"' IN BOOLEAN MODE);

Contents Haut

Installing MeCab From a Binary Distribution (Optional)

This section describes how to install mecab and mecab-ipadic from a binary distribution using a native package management utility. For example, on Fedora, you can use Yum to perform the installation:

yum mecab-devel

On Debian or Ubuntu, you can perform an APT installation:

apt-get install mecab
apt-get install mecab-ipadic

Contents Haut

Installing MeCab From Source (Optional)

If you want to build mecab and mecab-ipadic from source, basic installation steps are provided below. For additional information, refer to the MeCab documentation.