| The ISLE Project is a joint initiative supported by the European Community and the National Science Foundation of the USA. Following the EAGLES Project, ISLE aims at developing standards for natural language engineering, in three main directions: computational lexicons, multimodal human-machine interaction, and evaluation. The project has started in December 1999, the mid-term review took place in March 2001, and the end of the project is scheduled in March 2002. Members of ISSCO are involved in all of the three working groups (see below). The coordinator of the project is Professor Antonio Zampolli, from the Consorzio Pisa Richerche, Italy. |
|
CLWG - Computational Lexicon Working Group / Multilingual Lexicons |
| This group focuses on lexical semantics and multilinguality. Its goals are to extend the EAGLES guidelines and standards to bilingual/multilingual lexicons, to create a lexical entry tool as well as ISLE conformant samples, and to identify evaluation measures for lexicons. Pierrette Bouillon from ISSCO/ETI is involved in this working group. |
NIMM - Natural Interaction and Multimodality Working Group |
| This group is composed of several work packages, dealing with NIMM data resources, annotation schemata, systems and tools, metadata descriptions for MM resources. Andrei Popescu-Belis from ISSCO is involved in the work package on metadata standards (IMDI - ISLE Metadata Initiative). Coordinated by Peter Wittenburg (Max Planck Institute, Nijmegen, The Netherlands), this work package proposed a standard metadata set for language resources, encoded using XML, as well as tools for browsing and searching a collection of metadata. The workpackage is related to the OLAC initiative from the LDC (University of Pennsylvania). |
EWG - Evaluation Working Group: Machine Translation Evaluation |
| This group involves mainly researchers from ISSCO/ETI (Maghi
King, Sandra Manzi,
Andrei Popescu-Belis),
together with researchers from USC/ISI (Eduard Hovy, Elena Filatova), in
collaboration with The Mitre Corporation (Keith Miller, Florence Reeder).
The main focus of the working group is the development of a
classification or taxonomy of the features that are relevant to machine
translation evaluation. Based on previous attemps to evaluate machine
translation, the taxonomy organizes the various features hierarchically
and reviews extant metrics for each of them. In order to disseminate our
results and receive feedback, a series of four workshops is being organised:
Mexico, October 2000 (at AMTA 2000), Geneva, April 2001 (standalone hands-on
workshop), Pittsburgh, PA, June 2001 (at NAACL 2001), and Compostela, September
2001 (at MT Summit VIII). Once the methodology of such a taxonomy for evaluation
will be fully mastered, its extension to the other fields of language engineering
will be contemplated.
For more information about the Evaluation Working
Group |
Back to: ISSCO -- ISSCO's Projects