FGCS Assessment
Gilles Kahn
INRIA Sophia Antipolis, FRANCE
Introduction
Assessing the FGCS project is a hard task, not only because there is little time and
space to do so, but also because the scope of this project is very wide. I doubt if many
people can read equally competently the work that has been performed in so many
diverse areas: computer architecture, programming language design, database de-
sign, natural language analysis and generation, genome research. I for one have
very little competence on natural language work, although I hear from a variety
of sources that this may indeed be one of the strongest points of your work in
the last phase of the project.
The FGCS project is very broad, but everyone can see its unity of purpose. When
listening to all the presentations and reading some papers, I wonder whether there
has been enough time for true integration of the many components that have been
developed in the seven laboratories. I suggest the following problem, to see what I
have in mind. You have developed a theorem prover, MGTP. On the other hand, you
have sophisticated tools to generate sentences and paragraphs in Japanese. Assume
that you would like to connect these two components, so that the theorem prover
produces proofs in natural language, that a Japanese mathematician would like. In
particular, these proofs should not be too verbose, concentrate on the real difficulties
and be socially acceptable. Is this a problem that can be solved in a matter of weeks
or months with the software that you have developed, or do we need to start a new
project?
I have listened carefully to all the talks that reported the work of FGCS project,
and I must say that they were all very high quality presentations. The laboratory
chiefs show considerable experience, maturity in their scientific fields. Answers to
questions are very direct and frank and do not try to cover up difficulties when there
are some. As well, the demonstrations have shown strengths and weaknesses of the
software. I appreciate the considerable amount of work needed in preparing such a
thorough presentation of the FGCS achievements. This has confirmed an attitude that
I have witnessed in the wonderful INRIA-ICOT meetings that I have attended: a
completely frank exchange of views with scientists of high caliber, who are concentrat-
ing on basic research, and building software prototypes to demonstrate the validity of
their fundamental ideas.
- 67 -