Output Formatting

Appendix F: Output Formatting

MeCab's Default Output Format

MeCab's default ChaSen output format take the following form:

surface \t feature 

• surface is the morpheme itself
• feature is a comma-delimited string of the following elements:
  • part-of-speech
  • sub-class 1
  • sub-class 2
  • sub-class 3
  • inflection
  • conjugation
  • root-form
  • reading
  • pronunciation

Please refer to MeCab: Yet Another Part-of-Speech and Morphological Analyzer, とりあえず解析してみる.

Output Format Macros

It is possible to override the output format by customizing the Natto::MeCab node output format using the following macros.

Macro	Definition
%s	node stat status value: 0 normal, 1 unknown, 2 sentence start, 3 sentence end
%S	the input sentence
%L	length of input sentence (bytes)
%m	morpheme surface
%M	morpheme surface including leading whitespace (c.f. %pS)
%h	part-of-speech ID
%%	% char (escaped)
%c	word cost
%H	comma-delimited list of POS, conjugation, reading, etc.
%t	character type id
%P	marginal probability (only with -l2 option)
%pi	unique node ID
%pS	morpheme including any leading whitespace; same as %pS%m and %M
%ps	start position
%pe	end position
%pC	accumulative cost from previous node to this one
%pw	same as %c
%pc	accumulative cost + word cost (from sentence start)
%pn	accumulative cost + word cost (this morpheme only, %pw + %pC)
%pb	* for most optimal path; whitespace otherwise
%pP	marginal probability (only with -l2 option)
%pA	alpha, forward log probability (only with -l2 option)
%pB	beta, backward log probability (only with -l2 option)
%pl	length of morpheme (bytes), same as strlen (%m)
%pL	length of morpheme including any whitespace (bytes), same as strlen(%M))
%phl	left path id
%phr	right path id
%f[N]	Nth element of MeCab's default output feature
%f[N1,N2,N3...]	N1,N2,N3... elements of MeCab's default output feature, tab-separated
%FC[N1,N2,N3...]	N1,N2,N3... elements of MeCab's default output feature, delimited with char C; any whitespace elements are not output
\0 \a \b \t \n \v \f \r \\	the usual string formatters
\s	' ' (half-width whitespace)

Using the --*-format Options

You can define custom output formats using the above macros by using the --node-format, --unk-format, --bos-format, --eos-format or --eon-format options when instantiating Natto::MeCab.

Example 1: Specifying user-defined formats:

# pseudo-code, you would have to specify the output format macros as STR
# long-format style
nm = Natto::MeCab.new('--node-format=STR --bos-format=STR --eos-format=STR --unk-format=STR')

# short-format style
nm = Natto::MeCab.new('-F STR -B STR -E STR -U STR')

# Ruby hash
nm = Natto::MeCab.new(node-format: 'STR', bos-format: 'STR', eos-format: 'STR', unk-format: 'STR')

Using Configuration File mecabrc

It is also possible to define your custom output formats in the $MECAB_HOME/etc/mecabrc configuration file.

Example 2: Adding user-defined output formats to $MECAB_HOME/etc/mecabrc

# pseudo-code, you would have to specify the output format macros as STR
node-format-KEY = STR
unk-format-KEY = STR
eos-format-KEY = STR
bos-format-KEY = STR
eon-format-KEY = STR

Use the --output-format-type option to specify the user-defined output format KEY.

Example 3. Specifying user-defined output format KEY

# long-format style
nm = Natto::MeCab.new('--output-format-type=KEY')

# short-format style
nm = Natto::MeCab.new('-O KEY')

# Ruby hash
nm = Natto::MeCab.new(output_format_type: 'KEY')

Further details may be found at MeCab: Yet Another Part-of-Speech and Morphological Analyzer, 出力フォーマットの指定

---

Previous | Home | Next