1.  News

The VOODOO web frontend was down for a period of time. It is running again now, so feel free to try it out.

2.  Software Description

VOODOO is a software to detect the occurrence of category names in scientific texts. It is freely available under the GNU GPL. You can access VOODOO using its webinterface on http://onto.eva.mpg.de/VOODOO, or by getting the source from the CVS at our Savannah project page.

3.  Download

• http://onto.eva.mpg.de/voodoo/VoodooGUI.zip -- SWING user interface
• http://onto.eva.mpg.de/voodoo/VoodooConsole.zip -- Commandline interface

Please find the source in the CVS at our Savannah project page. You will need Java 6 to run VOODOO.

4.  Sample data

We have analyzed all fulltext articles of PLoS Genetics using VOODOO. The results of the analysis are here. For this analysis, we have used the Gene Ontology, the Disease Ontology and the Celltype Ontology. The tablefile used was http://onto.eva.mpg.de/voodoo/voodooanalysisCLEAR.tbl.

The whole PLoS Genetics corpus was analyzed in approximately 350 hours CPU time, most of which was needed for the parsing of the articles using the Stanford Parser. 39 articles were not completely analyzed due to a problem with the Stanford Parser. The command to run the analysis was java -Xmx2G -jar VoodooConsole.jar -l englishFactored.ser.gz -i infile.txt -o outfile.txt -t voodooanalysisCLEAR.tbl

5.  Tutorial

The VoodooTutorial gives a step-by-step introduction on how to use the Voodoo software.

6.  Feature Requests

• web interface:
  • option to upload files containing the text
  • write output to file and permit its download
  • list of jobs starting with the newest job
• command line interface:
  • parameter -l as default, only 3 parameters
  • output help when no arguments are given
• GUI
  • no language as input
  • write output to file