CCGlab

Lisp code for the masses, for something not so massive.

CCGlab is a tool to experiment with Combinatory Categorial Grammar (CCG).

CCGlab is designed with linguists, cognitive scientists, and computational linguists in mind. You can use it for quick check of CCG ideas, or for large-scale development of grammars and models.

We write CCGlab grammars as we do in CCG papers, and look at the output in almost paper format (ok, almost).

It implements all known combinators of CCG, and then some, in all variants and powers. Lexical rules, slash modalities, and meta-categories such as (X\X)/X have been implemented. The manual describes the range of possible CCGlab grammars.

It also implements probabilistic CCG, aka PCCG. In its tool-supplied form, only lexical features are used, which you can amend. Parameter estimation and update are provided. Therefore model training is possible. Models and grammars use the same format.

In deduction mode, CCGlab gives you all parses of a string, along with logical forms. In induction mode, it gives you the most likely LF for a string, its most likely derivation, and the most likely derivation for any LF for the string. Both are based on the same CKY parse engine and the same grammar format.

The main features of CCGlab which distinguish it from other publicly available CCG code:


CCGlab resources

Manual Latest manual. For reference, and for sneak preview.
Repository Github repo (with installer). Also contains sample grammars and release notes.
License Publicly available software, with GNU General Public License. One day it will be hardware.
Tips Practical information about writing .ccg grammars, working with projects, updates, etc.

disclaimer: We're not related to any real lab. We dwell in cyberspace, which is currently github.