Corpora: US company patent on common machine translation approach
Naomi Hallan (nhallan@pop2.restena.lu)
Wed, 15 Oct 1997 20:28:04 +0200
The EC (as it was then) ran an ESPRIT project on machine translation
called COMMUNAL in the 1980s. As I recall, it used Esperanto as the
intermediate between any pair of EC languages, parsers based on systemic
functional grammar, and content addressable file store for accessing the
knowledge base.
I have no idea whether any of the machine aids for translators currently
in use in the EU institutions actually use a system developed from this
project, nor do I know whether the groups involved in the project took
out patents on any parts of it, but certainly there must have been
reports on the project published (haven't had chance to check yet), and
I'm sure this puts it in the public domain.
Unfortunately I have no now contact to anyone who was involved (my
husband had a colleague who worked on the project, and he passed me some
documents to look at, but there aren't any names that look useful today),
but perhaps other list members do. I would imagine anyone whose
activities were threatened by this patent could get hold of information
from the ESPRIT programme. The little I know about it suggests that there
might well be similarities in technique as well as ideas to what Toltran
was claiming, which would mean they had no real basis to patent the
procedures.
Naomi