Textpresso is a text mining system for scientific literature whose capabilities go far beyond that of a simple keyword search engine. The two key elements are the collection of the full text of scientific articles split into individual sentences, and the implementation of semantic categories, for which a database of articles and individual sentences can be searched. The source of the full text articles are PDFs, and additional bibliographical information that is obtained from other citation databases can be processed as well.
Demo & Screenshots
Please visit the live main site at www.textpresso.org for examples and screenshots.
The package is designed for Linux operating systems and is tested to run on an Intel x86 based hardware. The required minimal disk space is around 6GB per 1000 full text papers, half of it is used by the publically (via WWW) accessible database, while the other half is needed for database preparation and maintenance. If necessary, the latter can be reduced.
- Software for a world wide web server such as Apache needs to be installed, and an Internet connection should exist
- Perl 5.6.1 or higher should be present, and the most common Perl packages.
- The installation script requires bash
- Mailer::Mail (in MailTools-1.58)
- PDF::Create (in PDF-Create).
- If the model organism database is based on ACeDB then AcePerl is required
- XPDF (http://www.foolabs.com/xpdf/), the pdftotext converter
- RBT, a part-of-speech tagger developed by Eric Brill (http://research.microsoft.com/~brill/)
RBT is distributed free of charge under a license of the Massachusetts Institute of Technology and the University of Pennsylvania. If you want to recompile either of the packages, you additionally need a C compiler.
Installation instruction can be found in the tarzipped package file and is called TextpressoManual.pdf.
A user guide is available online.
Hans-Michael Muller, mueller (at) caltech.edu