During 2012 we conducted a pilot project whose aim was to create the beginnings of a multilingual machine readable database building on Bill MacCartney's xml version of the FraCaS testsuite. Here are the translations and associated .dtd and .xsl files: Three of these languages use non-Latin writing systems and these will not work in all web browsers. They all work in Firefox.

We also did some experiments using the FraCaS Treebank to build inference rules based on GF abstract syntax representations.

If you find this useful or have comments please let me know ( We hope in the future to extend the test suite, translate to more languages, ground the inferences in real data, develop tools for analyzing the inferences and develop web-based facilities for collecting judgements on inferences. If you have done similar work or would like to be involved in future developments of this, let us know.

Last modified: Fri Oct 7 16:21:58 CEST 2016