Before you start

Before following any of these quickstart guides, be sure to read the JavaNLP Welcome for all the rules you're expected to abide by.

JavaNLP Quickstart

Register for an NLP account
- Email action@nlp.stanford.edu, requesting an account on the NLP machines and CC'ing your advisor.
Check out the code
- svn co svn+ssh://username@jacob.stanford.edu/u/nlp/svnroot/trunk/javanlp
Compile the code
- In ant:
  1. cd javanlp
  2. ant
- In Eclipse:
  1. File -> Import -> General -> Existing Projects into Workspace
  2. Select root directory: browse to the javanlp directory checked out by svn
  3. Projects: select JavaNLP-core, JavaNLP-research

Register for an NLP account (if you haven't already above)
Log in to jacob or jamie
- ssh username@jacob.stanford.edu or
- ssh username@jamie.stanford.edu
Set up your environment. Add the following to your .cshrc or .bashrc. (It must go at the end because it overrides things like umask.)
- source /u/nlp/bin/setup.csh
Select a machine to run your job on
- Visit the NLP Machines page
- Select a machine that's "free" or has "spare capacity"
- Never run big jobs on jacob, jamie, nlp or jay!
Start your job
- For example, if running a JavaNLP class from your javanlp checkout directory:
  - nohup bin/javanlp.sh edu.stanford.nlp.my.package.MyClass &
Keep an eye on your usage
- Check back at the NLP Machines page

Register for an NLP account (if you haven't already above)
Setup access to Linguistics corpora
- Follow the corpora access instructions from the Linguistics department
- Note that this involves emailing the Corpus TA with some very specific information, and you'll need to wait for a response before you can access the Linguistics department corpora
- Each time you login and want to access Linguistics corpora, you need to run:
  - /usr/kerberos/bin (and enter your SUNet password when prompted)
  - /usr/bin/aklog
Look around for your corpus
- /scr/nlp/data -- NLP corpora that don't need backed up (e.g. data from CDs we have)
- /u/nlp/data -- NLP corpora that need backed up (though in practice, other things are here too)
- /afs/ir/data/linguistic-data/ -- NLP and Linguistic corpora (mostly from LDC), as described on the inventory page
- If you search all these places can't find some corpus you want, feel free to ask on java-nlp-list@mailman.stanford.edu to see if anyone on the JavaNLP list knows if we have the corpus and where it might be.