Proceedings of the Workshop on Multilingual Language Resources and Interoperability - MLRI '06

Abstract

In an ever-expanding information society, many language processing systems are now facing the "multilingual challenge". Language resources, such as dictionaries, thesauri and wordnets, ontologies etc., as well as annotated corpora play an important role for the development, deployment, maintenance and exploitation of language processing systems. Much work on architectures for multilingual language resources, on recommendations of best practice for creating, representing, maintaining and upscaling such resources has been done in the 1990s, but since then, most efforts in this field have had less visibility. On the other hand, much research and development work has been done on techniques for acquisition of language data, on upper ontologies, on resource standardisation, and, last but not least, on the Semantic Web. One of the aims of this workshop it to provide an up-to-date view on issues relating to multilingual language resources and interoperability, in terms of language description, of technology and of applications. The development and management of multilingual language resources is a long-term activity in which collaboration among researchers is essential. We hope that this workshop will gather many researchers involved in such developments and will give them the opportunity to discuss, exchange, compare their approaches and strengthen their collaborations in the field. The impressive overall quality of the submissions (22) made the selection process quite difficult but we would like to acknowledge the dedication of our program committee who provided many useful comments to all papers. During the reviewing process we took the decision to accept only 9 papers (about 41%) in order to allow for more discussions during the workshop. The papers address a broad range of issues related with language resources for multilingual NLP applications, covering lexicons for general and specialised language, parallel corpora, and the acquisition of data from corpora. In particular, questions of lexical modelling and of standards for lexical resources, as well as approaches to interoperability and resource sharing in a distributed infrastructure are in focus. As multiwords are an important part of any practically usable lexical resource, two papers have been selected which deal with questions of the representation and the corpus-based acquisition of multiword items (here: collocations), from a multilingual perspective. Finally, techniques for detecting parallel texts (here: English/Japanese) and a new view on the Bible as a truly multi-lingual resource for cross-linguistic information retrieval will be discussed as examples of approaches to get access to new sources of data for the creation of language resources. Thus, the workshop covers central aspects of resource-related research; it is structured in a way to go upstream from lexicon standardisation and sharing, over lexical modelling to the identification and the use of corpora as a source of lexical data.