Ulrich Heid, Andreas Mengel
The MATE project aims to specify, develop and test an internet-based set of tools for the annotation of spoken dialogue corpora. Annotated dialogues are a central resource for the efficient development of speech and language technologies, including spoken language dialogue systems. The market for such systems is currently expanding and there is an increasing need for re-use of annotated material and for making the process of annotation more efficient.
This requires standardisation and the development of tools. The Natural Interactive Systems Group (NIS) at MIP (The Maersk Mc-Kinney Moller Institute for Production Technology, Odense University, Denmark) leads the MATE project, and will in addition mainly be responsible for communication problems mark-up and for specification of a workbench and implementation of its interface part which is expected to be done in Java.
Other partners in MATE are DFKI Saarbruecken, HCRC Edinburgh, ILC Pisa, IMS Stuttgart, CSELT Torino, TID Madrid, DFE Barcelona and Elsnet Utrecht.
At IMS, we are responsible for properties of the markup that enable the combination of different annotation levels (cross-level).