[Corpora-List] Open-Source Semantic Learning Software

Dominic Widdows dwiddows at csli.stanford.edu
Sat Feb 28 01:42:50 UTC 2004


Infomap NLP -- Open-Source Semantic Learning Software
-----------------------------------------------------

The Infomap Project at Stanford University's Center for the Study of
Language and Information (CSLI) is pleased to announce the release of
the Infomap NLP semantic learning software, version 0.8.3.

Please visit the Infomap NLP Semantic Learning Software home page,
http://infomap-nlp.sourceforge.net/

This software performs LSA-like processing on free-text corpora to
allow retrieval of words and documents similar to a query.  Special
features include:

    - A "NOT" function which removes one or several unwanted keywords, using
      the negation and negated disjunction connectives of quantum logic
      to remove unwanted senses of ambiguous words and unwanted meanings
      from retrieved documents.

    - An efficient C implementation capable of processing large
      corpora.  (We have used it extensively on the BNC, for
      instance.)

    - Allows for easy incorporation of part-of-speech tags and
      labelled multiword expressions (has been used with both).

    - Extensive documentation:
		- detailed man pages for the software's various
		component programs

		- a User Manual with a simple, clear introduction to
		get you started

		- an Implementation Guide for those interested in
		understanding and modifying the software

    - Highly permissive BSD license and full source-code access allow
      you to study the software and modify it as needed.

    - Project mailing lists for convenient reporting of bugs,
      communication with other users, participation in the development
      process, and receiving announcements of future releases.

We have been using this software for 3+ years, and it has contributed
to results in information retrieval, word sense disambiguation, and
lexical acquisition which are described in peer-reviewed, published
papers (also available from the website).  This is the first public
release.

Please visit our SourceForge homepage,
http://infomap-nlp.sourceforge.net/, or our SourceForge project
summary page, http://sourceforge.net/projects/infomap-nlp/, to
download the software.

We are eager to hear your feedback and to fix any bugs that you bring
to our attention.

                               Scott Cederberg & Dominic Widdows

--
Infomap Project
Computational Semantics Lab
Center for the Study of Language and Information (CSLI)
Stanford University

http://infomap.stanford.edu/



More information about the Corpora mailing list