Grammar Engineering for Deep Linguistic Processing (SS 2012)

Description

This course provides a hands-on introduction to the techniques and tools needed for building the precise, extensible grammars required both in research and in applications. Through a combination of lectures and in-class exercises, students will investigate the implementation of constraints in morphology, syntax, and semantics, working within the unification-based lexicalist framework of Head-driven Phrase Structure Grammar.

The course is heavily based on the open source DELPH-IN grammar engineering software tool-chain, including LKB, PET, [incr tsdb()], etc.

Course Info

Time:	Monday 14-16h
Office hours:	Thursday 11-13h (after email contact)
Location:	CIP Room (Except for the first meeting on 23.04 in U.15)
Type:	Project Seminar
Credits:	5 LP B.Sc./M.Sc.

Requirements

Participants should have studied grammar formalism or syntactic theory, and have a basic understanding of constraint-based grammar. Participants should also be familiar with basic *nix commands, text editing (emacs recommended), CVS handling, etc. Short tutorials will be given during the course if necessary.

The course is composed of a mixed series of lectures and lab sessions. Since this will be a highly interactive, hands-on course, active class participation and on-time submission of assignments will be viewed favorable when it comes to grading.

Schedule

Exercise Grammars

• grammar01
• grammar02
• grammar03
• grammar04

Software Setup

LKB

Before starting LKB, make sure you have a tmp directory under your home directory. LKB uses it for caching large grammar lexicon:

 $ mkdir -p ~/tmp

To run LKB locally on a COLI machine:

 $ export DELPHINHOME=/proj/delphin
 $ emacs -l /proj/delphin/lkb/etc/dot.emacs

Then, in the emacs:

 M-x lkb

To run LKB remotely on a COLI machine, you need a working X-window server on your machine, and SSH access to the COLI network.

 $ ssh -Y login.coli.uni-saarland.de
 $ ssh -Y cluster-1.coli.uni-saarland.de
 $ export DELPHINHOME=/proj/delphin
 $ emacs -l /proj/delphin/lkb/etc/dot.emacs

Of course, you can substitute cluster-1 with any other available cluster node. Then, in the emacs:

 M-x lkb

Be sure to use -Y or -X option to enable X11 forwarding if you are running LKB remotely.

Alternatively, you may install binary builds of LKB on your local machine. Currently, Linux x86_32 and x86_64 platforms are supported with an automatic installation script. Binary builds for Solaris, Windows and MacOS (PPC-based) are also available for older versions of LKB. For Linux platform installation:

 $ mkdir ~/delphin
 $ export DELPHINHOME=~/delphin
 $ wget http://lingo.stanford.edu/ftp/etc/install
 $ chmod +x install
 $ ./install

Then follow the instructions from the install script. In case of any problem, please contact the lecturer.

Grammar Matrix Type Hierarchy

References

[Copestake et al., 2005]

Ann Copestake, Dan Flickinger, Carl J. Pollard, and Ivan A. Sag. Minimal recursion semantics: an introduction. Research on Language and Computation, 3(4):281–332, 2005.

[Sag et al., 2003]

Ivan Sag, Thomas Wason and Emily Bender. Syntactic Theory: A Formal Introduction, Second Edition. CSLI, Stanford, USA, 2003.

[Copestake, 2002]

Ann Copestake. Implementing Typed Feature Structure Grammars. CSLI, Stanford, USA, 2002.

[Bender et al., 2002]

Emily Bender, Dan Flickinger, and Stephan Oepen. The Grammar Matrix: an open-source starter-kit for the rapid development of cross-linguistically consistent broad-coverage precision grammars. In Proceedings of the Workshop on Grammar Engineering and Evaluation at the 19th International Conference on Computational Linguistics, pages 8–14, Taipei, Taiwan, 2002.

[Krieger and Schäfer, 1994]

Hans-Ulrich Krieger and Ulrich Schäfer. TDL - a Type Description Language for HPSG. Technical Report RR-94-37, Deutsches Forschungszentrum für Künstliche Intelligenz GmbH, 1994.

[Carpenter, 1992]

Bob Carpenter. The Logic of Typed Feature Structures. Cambridge University Press, Cambridge, UK, 1992.

---

Last modified: May 21 2012.