UPDATE - 11/29/06

I've added tarballs for easy downloading:

everything.tgz  - contains everything in the directory
neededfiles.tgz - contains the files that you need for the assignment

---

You don't need to run prepare.sh; all that has already been done
and the scripts are included for completeness.  You can proceed
immediately to loading things into MATLAB and deriving the EM
updates.

Once you are finished inserting the proper lines in ibm2_train_lm.m
and ibm2_train.m, you should be able to run a session like this:

% almost instantaneous
delex = lexicon('data-de');
% almost a minute
enlex = lexicon('data-en');
% may take several minutes (but has status indicator)
[english,deutsch,lmenglish]=europarl();
% may also take several minutes (with status indicator)
[N,mmax]=size(english); mmax=size(lmenglish,2)-mmax;
[LM,LMc]=ibm2_train_lm([lmenglish;english,sparse(N,mmax)]);
% should be very quick
[T,D,lom]=ibm2_train(english,deutsch);
% also very quick
germans = klaus();
for i=1:size(germans,1)
  german = germans(i,:)
  englisch = ibm2_beam_decoder(T,D,lom,LM,german)
  numtostr(delex,german)
  numtostr(enlex,englisch)
end
