The project uses Eckhard Bick and Tino Dideriksens vislcg3, a compiler for constraint grammar, a formalism originally developed by a.o. Fred Karlsson and Pasi Tapanainen at the University of Helsinki.
So far, we have a disambiguation file sme-dis.rle, for North Sámi, and a file smj-dis.rle for Lule Sámi. In addition to that, we are working on beta versions for Greenlandic and Faroese. The content of the North Saami file is documented here.
Other versions of constraint grammar
The philosophy behind constraint grammar is relatively simple, and the programming algorithms are found in several different version (the version we use at giellatekno is no. 5 on this list).
- The original CG formalism was developed by Fred Karlsson, at Dept. of Linguistics in Helsinki
- Bart Jongejan's C++ version of the original CG
- A new version of CG, called CG-2, was developed by Pasi Tapanainen, cf. Tapanainen 1996. It is maintained by Connexor
- vislcg, an open source implementation of CG-2, at http://visl.hum.sdu.dk/visl/. This parser is available for download ad sourceforge, here.
- vislcg3, a new generation version ov the open source vislcg is currently being developed at http://visl.hum.sdu.dk/visl/.
Other disambiguation tools
Grammar-based disambiguation tools
- disamb, at Xerox
- FSIG (Finite-state intersection grammar), at Dept. of Linguistics in Helsinki
- LFG and HPSG-based tools ??
For various reasons, do not use these tools, but have chosen constraint grammar instead.
Statistically-based disambiguation tools
This is the dominating approach to disambiguation. Here, we think the grammar-based approach is on the right track, and we thus do not use statistically-based tools. In the future it still may be relevant to have a look at what statixtically-based tools may achieve when applied on the top of grammar-based disambiguation tools, in order to resolve on statistical grounds such open cases that cannot be dealt with by grammatical tools alone.
Thus, we should at some point start orientating ourself in the world of statistically-based disambiguation tools.
Links to old documentation
These links are kept for reference, but not relevant for present usage.
A disambiguation bibliography
Last modified: $Date: 2012-11-22 13:41:43 +0100 (tor, 22 nov 2012) $, by $Author: trond $