Karel Oliva (oliva@coli.uni-sb.de)
Mon, 8 Dec 1997 22:47:37 +0200
Hi, with all respect to Shalom's opinion, I have to say, regretfully, that I have to side with Tibor ("regretfully" because I usually NEVER side with Tibor :-) ) First of and above all: the discussion of this topic should have been performed BEFORE the decision was taken - it is maybe nice to discuss it now, but I am afraid it is no more helpful. But to come to the point: as for (as Shalom puts it) "the audience consist(ing) of formal and computational linguists who are open to our approach", I have to say that following some funny experience I am somehow less enthusiastic. To make the point, I have to say, first, that it is a very personal experience. So whoever thinks anything like this does not belong here (which I suspect is the case), should delete this mail NOW ! For those who risked not to do so, I have to say (second) that Drew Moshier, Sabine Lehmann and myself proposed a paper to FG 97, and got the review which - for fun, but maybe also for more - I append below. With (once more) all respect to the freedom of speech of the reviewers (and to the fact that the individual reviewing was blind - but hardly the final decision of the chairperson(s) of the Program Committee to forward such a review), it seems to me to be rather exaggerated a statement to say that Drew does not understand the technical perspective of HPSG and that his statements about feature structures are patently false ... or, in the case it is not exaggerated, it seems to be a proof of the level of openness of the FG people to our approach :-) But, in any case, I understand that we cannot change it now, so take it as it is (but be forewarned). karel oliva ========================================= Abstract 32 What's good: very pretty pictures :-) What's bad: a bit vague and almost entirely incomprehensible from a technical perspective. not clear author really understood HPSG, because some statements are patently false (ie. feature structures can't have non-trivial intersections -- they can in HPSG, no matter what is meant by 'interesection' -- unification or generalization). Author should read Moshier's published work on this same topic (in the last couple of years of JoLLI). ========================================== N.B. I particularly liked the logic of the wording "(it is patently false that) feature structures can't have non-trivial intersections -- ... no matter what is meant by 'interesection' " With other words, and admitting fully that the proposed paper _was_ weak (and hence this is NOT a complaint(*)), this quotation seems to me to be a pointer which should make us think what the FG people expect to appear and how - under some circumstances - they might be going to review the rest ... (*) even when I am sure I am not going to convince anybody that it is not ... but this is another point I cannot change :-)
This archive was generated by hypermail 2.0b3 on Fri Dec 18 1998 - 20:36:10 PST