LINGUIST List 13.1366

Wed May 15 2002


Editor for this issue: Terence Langendoen <>


  1. Terry Langendoen, AVAILABLE FOR REVIEW


Date: Wed, 15 May 2002 22:42:47 -0400 (EDT)
From: Terry Langendoen <>

The LINGUIST list is seeking two reviewers for the software
package release described below (following the line of asterisks).

If after reading the description of this software package you
decide that you are interested in reviewing it, please send an
email message to that effect to Terry Langendoen at

Your message should contain the following information.

1. Your full name and email address.
2. A statement of your qualifications to do the review, including
 access to a computer on which you can run the software

Please do not provide a URL where you describe your qualifications
or include that information in an attachment. Send the information
as plain text in the body of your message.

If you are selected to do a review of this software, Terry will send
you instructions for accessing the software. Your review is due within
six weeks of your selection as a reviewer. It will be posted as a
"Reviews" issue.
VisualText 1.2, by Text Analysis International, Inc.

VisualText is the first integrated development environment (IDE) for
natural language processing (NLP). It is the most advanced commercial
tool available for constructing not only information extraction and text
analysis systems, but the whole gamut of NLP applications, such as
machine translation, language generation, ontology building,
categorization, summarization, grammar critiquing, and more.

VisualText builds multi-pass, multi-strategy analyzers that communicate
via a single parse tree per text and also via a shared hierarchical KBMS
(Knowledge Base Management System) called Conceptual Grammar.
Constraining the analysis to produce a single parse tree practically
forces an efficient and fast analyzer to be built. Ambiguity
information and competing hypotheses can be carried in the parse tree

A key aspect of VisualText is a rich, highly integrated user interface
that ties together text views, parse tree views, knowledge base views,
highlighting of text corresponding to the work done by each pass,
instant navigation from syntax errors and parse tree nodes to their
corresponding rule and line of code, and much more.

VisualText incorporates the simple yet powerful NLP++ programming
language, which combines rewrite rules, a C/C++ style programming
language, and data types for parse tree nodes, parse tree semantics, as
well as knowledge base concepts, attributes, and values.

The VisualText development environment interprets NLP++, supporting a
rapid modify-and-test development cycle. An automated rule generation
capability applies to user-highlighted (or "annotated") text samples,
supporting the construction of analyzers that can be maintained by
non-expert users.

The development environment is supported on Windows NT, 2000,
98, and ME. The runtime libraries for analyzers developed (and
optionally compiled) with VisualText run on Windows, Linux, and
effectively on any platform with a Gnu C++ compiler or equivalent.
VisualText is an open architecture, allowing user or 3rd party code to
be integrated with an analyzer, as well as supporting user extensions to
the NLP++ language itself.

To enable faster implementation of analyzers, TAI can make available
prior to release two general analyzers, TAIParse and GenTxt. Sample
applications include an advanced analyzer for web resumes.

VisualText can serve not only as a development environment, but as a
workbench for analysts, for annotation, and quality review. It is an
ideal substitute for languages such as Perl in many document processing,
format conversion, and pattern matching tasks.

Documentation is extensive and well-integrated with the development

Academic licensing is available to students, teachers, and academic
organizations. A 30-day trial version is available. For further
details, see

or contact us at


USA toll-free number: 1-877-235-6259
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue