Fadd: Finite-State Library for Alpino Grammar

What is it?

Fadd library is a collection of functions for using finite-state automata in Natural Language Processing. It is based on the fsa package, and it uses programs from that package for its operation. It has been created for the Alpino Grammar system in the framework of the Algorithms for Linguistic Processing, an NWO PIONIER project. The library contains functions for:

Using compressed language models, such as bigrams, trigrams, etc.;
Performing morphological analysis;
Restoring missing diacritics;
Performing perfect hashing.

Availability

The library is available under LGPL at ftp://ftp.pg.gda.pl/pub/software/xtras-PG/fsa/fadd.tar.gz. The current version is 0.14. From version 0.10 on, fadd supports sparse matrix representation in addition to the list representation. This brings it in sync with my fsa package.

Documentation

File README in the distribution contains (hopefully) usefull information on how to use the library; there are also some man pages. Compressed language models have been described in:

Jan Daciuk, Gertjan van Noord, Finite Automata for Compact Representation of Language Models in NLP, Sixth International Conference on Implementation and Application of Automata, CIAA'2001, Pretoria, South Africa, July 23-25, 2001.

The library was presented at CLIN 2001 in a talk:

Jan Daciuk, Gertjan van Noord, A finite-state library for NLP, CLIN 2001, University of Twente, Enschede, the Netherlands, November 2001.

Jan Daciuk