BioLG is a modification of the Link Grammar Parser of Sleator and Temperley, adapted for the biomedical domain as described in Lexical Adaptation of Link Grammar to the Biomedical Sublanguage: a Comparative Evaluation of Three Approaches (Sampo Pyysalo, Tapio Salakoski, Sophie Aubin and Adeline Nazarenko; BMC Bioinformatics 2006). If you publish work using BioLG, please cite this paper!
The latest release of BioLG with full source code can be downloaded here: biolg-1.1.12.tar.gz. The software is under the GPL-compatible Link Grammar licence. The included PCRE regular expression library is under the BSD licence, and the expat XML parsing library is under the MIT/X Consortium license. See the included LICENCE files for further details.
- Unpack the package: tar xvzf biolg-1.1.12.tar.gz
- In the biolg-1.1.12/pcre-5.0 subdirectory, run ./configure
- In the biolg-1.1.12/expat-2.0.0 subdirectory, run ./configure
- In the biolg-1.1.12 directory, run make
Please address any questions to sampo.pyysalo at it.utu.fi
BioLG retains the features of Link Grammar Parser, and its thorough documentation, found on the Link Grammar documentation page, applies also to BioLG.
BioLG features a number of extensions not included in the original Link Grammar. These include XML input and output modes, a regular expression-based system for specifying morpho-guessing rules (an explanation of morpho-guessing can be found in the Introduction to the Link Grammar Parser), the possibility to specify part-of-speech tags as part of the input, and support for specifying multi-word terms in the input (XML input mode only).
BioLG features easily extendable morpho-guessing rules, described on the morpho-guessing page.
Part-of-Speech tag input
BioLG can resolve unknown words based on their Part-of-Speech tags, as described on the page on Part-of-speech tag input.
XML input and output
Term input (beta)
BioLG features support for multi-word terms (beta), described on the page on term support.
Adapted Grammar (beta)
A beta version of the BioLG grammar, adapted to the biomedical domain, is now available on the Grammar adaptation page.
The names "Link Grammar" and "Link Parser" are used with permission of Daniel Sleator.
This page is under construction. More detailed instructions and description of the software to follow soon.