Install Oslo Bergen Tagger - presemt-ntnu/transglobal GitHub Wiki

This step can be skipped if you have downloaded a copy of the local data and do not want to build to your own local data

The Oslo-Bergen Tagger (OBT) is required for tagging and lemmatizing Norwegian input. Note that this is currently not well supported.

1. Install vislcg3 following the instructions on http://beta.visl.sdu.dk/cg3/chunked/installation.html#macosx. Building all dependencies takes a long time...

2. Find the run.sh command line script from the PRESEMT subversion repository. It is in PRESEMT/tools/OBT/osx/run.sh. Copy runs.sh to a new file obt and change

./vislcg3

into

vislcg3

so the version in /opt/local/bin/vislcg3 will be used.

3. Check that obt produces output:

$ echo "Jeg prøver lykken." |/Users/erwin/Projects/Presemt/svn/PRESEMT/tools/OBT/osx/obt
Warning: Line 5364: We don't think mixing NOT and ** makes sense...
Warning: Line 5373: We don't think mixing NOT and ** makes sense...
Warning: Line 18819: We don't think mixing NOT and ** makes sense...
Warning: Line 18820: We don't think mixing NOT and ** makes sense...
Warning: Line 23917: We don't think mixing NOT and ** makes sense...


model loaded
tagger compiled



Jeg	jeg	pron_ent_pers_hum_nom_1
prøver	prøve	verb_pres
lykken	lykke	subst_appell_mask_be_ent
.	$.	<punkt>

(Ignore the warnings)

3. Specify the full path to the obt executable in the env/tg-default.cfg file like this:

[tagger] # tagger/lemmatizer
    ...
    
    [[no]]
    command = /Users/erwin/Projects/Presemt/svn/PRESEMT/tools/OBT/osx/obt

It would be nice if obt could be made available through a symlink or an alias, but there seem to be technical problems with that.