Main Leopar page

Resources developer's documentation

If you want to develop your own resources for the Leopar parser, you will have to produce several linguistics resources. Leopar's resources are built with the Interaction Grammar (IG) formalism. A IG-grammar is described by a list of polarized tree descriptions.

WARNING: In the following, there are two notions of “grammar”:

  • the anchored grammar in which PTD are associated to words
  • the unanchored grammar in which PTD are not associated to words

The anchored grammar is used as the input of the parsing but it is too big to be described extensionally and so it is dynamically constructed from an unanchored grammar and a lexicon. Leopar contains an internal mechanism to combine the grammar and the lexicon.

In the following, we use the term grammar to speak about the unanchored grammar.

You can see an example of a lexicon entry, a grammar entry and the anchoring process on the page Overview of the parsing process.

The lexicon

To define your lexicon, you need:

The grammar

Leopar reads grammar in an XML format (see the DTD). Except for very small grammar, you should use the XMG meta-grammar compiler. To install XMG, you have to follow instruction on the page XMG2.

Hello World

Itig is a VERY small set of Italian resources for Leopar. It is a kind of “Hello World”, i.e. the smallest set of resources to begin with the development of IG resources.