The QALL-ME benchmark is a collection of several thousand spoken utterances related to the domain of tourism, both audio files and their corresponding transcriptions, in the four languages involved in the project: English, German, Italian and Spanish. These utterances ask for information about cultural events, accommodation, movies, gastro, etc. and have been transcribed according to guidelines set out by the QALL-ME consortium. The German, Italian and Spanish questions are also translated into English. Annotation of speech acts and expected answer type (EAT) have been completed. The benchmark is a complete, reusable multilingual resource which can be used for training and testing in the field of QA.
The QALL-ME benchmark consists of 16722 utterances in the four languages English, German, Italian and Spanish. [...read more]
Speakers in the QALL-ME benchmark
Utterrance duration in the QALL-ME benchmark
Read and created questions in the QALL-ME benchmark
Update: September 2009 [pdf, 37 KB]
Cabrio E., Kouylekov M., Magnini B., Negri M., Hasler L., Orasan C., Tomas D., Vicedo J. L., Neumann G., Weber C., (2008), The QALL-ME Benchmark: a Multilingual Resource of Annotated Spoken Requests for Question Answering, in Proceedings of the Language Resources Evaluation Conference (LREC), Marrakech, Morocco, May. [pdf, 118 KB]
The QALL-ME benchmark is distributed under Creative Commons Licence 3.0 (Attribution-Noncommercial-Share Alike 3.0 Unported).