The Oslo-Bergen Tagger is a morphosyntactic tagger for Norwegian bokmål and nynorsk. For general information about the tagger, visit its home page: http://www.tekstlab.uio.no/obt-ny/.
Installation and usage
The tagger consists of three parts:
- A multitagger (tokenizer, morphological analyzer, and compund analyzer)
- A Constraint Grammar (CG) tagger
- A statistical tagger (currently only for bokmål)
Watch the installation on YouTube (On OS X Yosemite)
The multitagger is currently only distributed in binary form. Compiled binaries for
32- and 64-bits Linux and 64-bits Mac OS X can be downloaded
from our server at the Text Laboratory, University of
http://www.tekstlab.uio.no/mtag/osx64/mtag-osx64, respectively). The
file should be placed in the
bin directory. If necessary, rename the file to
and make it executable. For example:
$ cd The-Oslo-Bergen-Tagger/bin $ wget http://www.tekstlab.uio.no/mtag/osx64/mtag-osx64 $ mv mtag-osx64 mtag $ chmod +x mtag
The Constraint Grammar tagger
Check out the VISL CG-3 repository from the Subversion repository at the University of Southern Denmark and install it. The repository can be checked out anywhere on your machine since it will be installed into a central location such as
/usr/local/bin. Installation instructions for various platforms can be found at http://beta.visl.sdu.dk/cg3/chunked/installation.html.
CG rules for morphological disambiguation of bokmål and nynorsk are found in the
nn_morf.cgshould be used when you only want to do CG tagging of bokmål and nynorsk, respectively. The CG tagger may leave some ambiguity, either because it is not confident enough to do complete disambiguation, or because there is genuine ambiguity in the material (such as nouns that can be analyzed as either masculine or feminine).
bm_morf-prestat.cgshould be used when you want to run statistical disambiguation after CG disambiguation in order to obtain completely disambiguated output (currently only available for bokmål). This is useful, for instance, for many language technology purposes. On the other hand, it may remove genuine ambiguity from the text.
The statistical tagger
Clone the OBT-Stat git repository from GitHub in the root folder of the distribution:
$ git clone git://github.com/andrely/OBT-Stat.git
Running the tagger
Shell scripts are included which will run the entire process: multitagging, CG disambiguation, and optionally statistical disambiguation (for bokmål).
CG and statistical disambiguation, bokmål:
$ ./tag-bm.sh TEXTFILE > DISAMBIGUATED_OUTPUT_FILE
CG disambiguation only, bokmål:
$ ./tag-nostat-bm.sh TEXTFILE > DISAMBIGUATED_OUTPUT_FILE
CG disambiguation only, nynorsk:
$ ./tag-nostat-nn.sh TEXTFILE > DISAMBIGUATED_OUTPUT_FILE