Introduction

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)

The multitagger

The multitagger is currently only distributed in binary form (source code will be available soon). Compiled binaries for 32- and 64-bits Linux can be downloaded from our server at the Text Laboratory, University of Oslo (http://www.tekstlab.uio.no/mtag/linux32/mtag32 and http://www.tekstlab.uio.no/mtag/linux64/mtag, respectively). The file should be placed in the bin directory. For example:

$ cd bin
$ wget http://www.tekstlab.uio.no/mtag/linux64/mtag

The Constraint Grammar tagger

1. 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.

    $ svn co --username anonymous --password anonymous
        http://beta.visl.sdu.dk/svn/visl/tools/vislcg3/trunk vislcg3
    $ cd vislcg3/
    $ ./autogen.sh
    $ make
    $ ./test/runall.pl
    ... and if all tests succeed ...
    $ make install
   

   If you encounter any problems, have a look at the installation instructions at http://beta.visl.sdu.dk/cg3/chunked/installation.html

2. CG rules for morphological disambiguation of bokmål and nynorsk are found in the cg folder.

   bm_morf.cg and nn_morf.cg should 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.cg should 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

1. Clone the OBT-Stat git repository from GitHub in the root folder of the distribution:

    $ git clone git://github.com/andrely/OBT-Stat.git
   

2. Download the HunPos distribution for your operating system and unpack it inside the OBT-Stat/hunpos folder (producing a folder called, e.g., OBT-Stat/hunpos/hunpos-1.0-linux). HunPos can be downloaded from http://code.google.com/p/hunpos/downloads/list. For example:

    $ cd OBT-Stat/hunpos
    $ wget http://hunpos.googlecode.com/files/hunpos-1.0-linux.tgz
    $ tar xvzf hunpos-1.0-linux.tgz 
   

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