TuLiPA and the lexicon (1)

Grammar Implementation with TAG

TuLiPA and the lexicon (1)

29. June 2011

Timm Lichte

Setting up the system on your account

TuLiPA is part of the VM xmg&tulipa, that needs to be startet with VirtualBox.

Start the VM .
At some point a login screen appears: user name = xmguser, password = xmguser
Open a terminal and type:
```
$ cd tulipa/toygrammar
$ ls 
```

You will see a metagrammar (spread over several files), a lemma lexicon (lemma), and a morphological lexicon (morph).

Compiling the source code

To compile the metagrammar type:

$ MetaTAG valuation.xmg --xml -o toygrammar.xml

To compile lemma type:

$ lexConverter  -tL -i lemma -o lemma.xml

To compile morph type:

$ lexConverter -tM -i morph -o morph.xml

The compiled files are regular XML-files, hence each has the postfix "xml".

Starting and using TuLiPA

There are two ways to start TuLiPA:

Start it with the terminal:
```
$ java -jar ../TuLiPA-2.0.2.jar
```
Start it by clicking the launcher icon in the top panel.

In either case, a GUI should appear where you can specify the location of the compiled resources.
Note that "axiom" should be set to "s" (the category of the root node)!

Now try to parse the following sentence:

(1) Peter loves Mary

Exercises

Extend the metagrammar, the lemma lexicon, and the morphological lexicon, such that they cover the follwing sentences:

(1) John sleeps
(2) the kids sleep
(3) every boy sleeps
(4) who likes John
(5) what does John like
(6) whom does Mary like
(7) Mary always sleeps
(8) John seems to sleep
(9) John tries to sleep
(10) Mary gives John a book
(11) Mary gives a book to John
(12) To John Mary gives a book
(13) what does John think Mary likes

Timm Lichte

Last modified: Tue June 28 12:00:00 CEST 2011