The Météo bitexts

This page gives access to two bitexts (Report and Warning) we created at RALI while working on the METEO Project described in the following publications.

Context of the work

In 2004, Rick Jones and Marc Besner from the Canadian Meteorological Centre of Environment Canada provided us with files containing both French and English weather forecasts produced during 2002 and 2003 (from these we created the Report bitext). The source report was usually in English, and its translation was produced either by a human or by the current METEO system.

In 2005, they provided us with files containing weather alerts issued over 2000-2004 (from these we created the Warning bitext). Given the special nature of this information for the public, the translation of this material is done by professional translators of the Canadian Translation Bureau.

In both cases, we received archived versions of the forecasts, not their original as they are published daily on the web, where the text is capitalized and for which the French translation (or text) does not contain any accent or diacritics.

The Report bitext

During his master thesis, Thomas Leplus organized this material into a bitext. The details of this operation -- which turned out to be more complicated than we initially envisioned -- can be found in his master thesis. The resulting bitext is organized into files (one per month), the format of which should be easily understandable by inspection.

The Warning bitext

Simona Gandrabur prepared the bitext from the raw material we originally received. Three types of alerts are differenciated: omnibus, violent storm and tornado warnings.

Tools

XML resources for manipulating meteo files (DTD, XML Schema, XSL stylesheet and a CGI for displaying XML in HTML (not meteo specific).

Credits

We are happy to share this data with you. We would appreciate that you cite this reference in any communication that would result from this material:

Philippe Langlais, Simona Gandrabur, Thomas Leplus and Guy Lapalme, "The Long-Term Forecast for Weather Bulletin Translation", Machine Translation, Kluwer Academic Press, 2005.

Note that the sentence alignment has been carried out by our sentence aligner Japa therefore expect some errors...

In any case, do not hesitate to contact Philippe Langlais.