home

QA@CLEF-2004

Test Sets

We considered 7 target languages, proposing 56 tasks (monolingual and cross-language).
Tasks are defined according to a source language (i.e. the language of the questions) and a target language (i.e. the language of both document collection and answers). In the monolingual tasks, source and target languages are the same.
Bulgarian and Finnish are just source languages (the latter was included exclusively against the English document collection).
The monolingual English task was not enacted in 2004, but since some participants would like to perform error analysis in order to evaluate the machine translation components they used in the tasks with English as target, the EN=>EN test set is available, as well.

Using the form below you can select the languages of each task and download the corresponding test set. Test sets are plain text files, UTF-8 encoded. For further information about the format, please read the guidelines we posted.