Second
Workshop on Example-based Machine Translation
Introduction
This is the second workshop on Example-based Machine Translation (EBMT)
of its kind to be hosted by the MT Summit X, 2005 in Phuket, Thailand. The
1st EBMT workshop took place in 2001 at the MT Summit VIII in Santiago de
Compostela, Spain. While the first workshop resulted in a book "Recent Advances
in Example-Based Machine Translation" (2003) which summarises the numerous
techniques used in our field and helped bring EBMT to a new audience, we
also hope to be able to publish these workshop proceedings such that it is
available to a wider audience.
Four years after the first EBMT workshop, the field has considerably
matured and evolved. In this second workshop we see a continuation of previous
systems and approaches as well as a number of new and innovative methods
and applications. We have interesting papers on semantic and type-driven
approaches to EBMT, approaches that look beyond the sentence border, retrieval
of fragments, example-based sign-language translation, corpus-based generation,
and investigations on data assembly and corpus consistency. Projects such
as METIS investigate example-based methods to machine translation that make
use of a monolingual TL corpus. In addition to these 'new' horisons, the
workshop also presents progression of work and approaches alread known from
the the previous workshop.
We have excellent papers on tree-based approaches, pure as well as template-driven
EBMT, and as in the 2001 workshop, position papers on what actually constitutes
an EBMT system. In general we observe a continuing desire for the integration
of various techniques and and a strengthening of the statistical underpinning
on which EBMT is based. Given the variety of different topics and methods
assembled in this workshop we find that research in EBMT is vibrant and
catalyzes an active research community trying to integrate and make sense
out of the various corpus-driven approaches to MT. In particular, as co-organisers
of this 2nd workshop, we are delighted to welcome a number of researchers
who haven't previously published in the area of EBMT. We intend to close
the workshop by a panel session where the contributors are asked to envisage
what EBMT will be like in the future, and what research directions might be
foreseen in the time between now and the 3rd EBMT workshop, and beyond.
We trust that you enjoy this workshop. We have enjoyed putting the programme
together, and we wish to offer many heartfelt thanks to our assembled Programme
Committee, each of whom did sterling work over and above what might reaonable
have been expected from them. The quality of the papers received was very
high, and if possible the quality of the reviews even higher!
Andy & Michael.