|
Size: 820
Comment:
|
Size: 1521
Comment:
|
| Deletions are marked like this. | Additions are marked like this. |
| Line 11: | Line 11: |
| See docs directory in the Segment package file. | For detailed info see the docs directory in the Segment package file. |
| Line 13: | Line 13: |
| == Using as sentence splitter == | === Using as sentence splitter for Polish === |
| Line 15: | Line 15: |
| In segment main directory type: {{{./bin/segment -l <language_code> -s <SRX_file>}}} For example:<<BR>> |
In segment main directory: |
| Line 23: | Line 19: |
| This way the program will read data from stdin and write sentences to stdout - each one in separate line. | This way the program will read text from stdin and write sentences to stdout - each one in separate line. |
| Line 25: | Line 21: |
| Recommended SRX rules file for sentence splitting in Polish written by Marcin Miłkowski: [[attachment:sample.srx]]. | Note: the SRX file can deal with text that contains paragraphs separated with two line breaks, and where a single line break is still inside the paragraph (which is also the default TeX mode). To have such behavior, use: {{{./bin/segment -l pl_two -s sample.srx}}} Adding an end-of-sentence marker on a single line break is achieved this way: {{{./bin/segment -l pl_one -s sample.srx}}} ==== Recommended resources: ==== * SRX rules file for sentence splitting in Polish, written by Marcin Miłkowski: http://languagetool.svn.sourceforge.net/viewvc/languagetool/trunk/JLanguageTool/src/resource/segment.srx. * More information about SRX format: http://morfologik.blogspot.com/2009/11/talking-about-srx-in-lt-during-ltc.html * Some notes on sentence splitting performance (in Polish): http://morfologik.blogspot.com/2010/03/testy-segmentacji.html |
Segment
Segment program is used to split text into segments (sentences, paragraphs, words). Split rules are read from file in XML based Segmentation Rules Exchange (SRX) standard format. Can be used as a programming library.
Homepage: http://sourceforge.net/projects/segment/
Documentation
For detailed info see the docs directory in the Segment package file.
Using as sentence splitter for Polish
In segment main directory:
./bin/segment -l pl -s sample.srx
This way the program will read text from stdin and write sentences to stdout - each one in separate line.
Note: the SRX file can deal with text that contains paragraphs separated with two line breaks, and where a single line break is still inside the paragraph (which is also the default TeX mode). To have such behavior, use:
./bin/segment -l pl_two -s sample.srx
Adding an end-of-sentence marker on a single line break is achieved this way:
./bin/segment -l pl_one -s sample.srx
Recommended resources:
SRX rules file for sentence splitting in Polish, written by Marcin Miłkowski: http://languagetool.svn.sourceforge.net/viewvc/languagetool/trunk/JLanguageTool/src/resource/segment.srx.
More information about SRX format: http://morfologik.blogspot.com/2009/11/talking-about-srx-in-lt-during-ltc.html
Some notes on sentence splitting performance (in Polish): http://morfologik.blogspot.com/2010/03/testy-segmentacji.html