<div dir="ltr"><div><div><div><div><div><div><div><div>G'day,<br><br></div>another difference between ACE and the LKB is in the generation of underspecified properties. In this case, we make substantial use of the LKB's behaviour, and would prefer ACE to behave in the same way (possibly with an optional choice).<br><br></div>The question is one of what happens when you specify a type in the middle of a hierarchy.<br><br></div>Consider the sense 'PRON sleeps'<br></div>If PRON has PERS: 3 NUM: sg and no gender, LKB and ACE both generate <br></div>He sleeps<br></div>She sleeps<br></div>It sleeps<br></div><div>They sleep. [androgynous they]<br><br><br></div>If we restrict the gender to m-or-f [PERS: 3 NUM: sg GEN: m-or-f]<br><div><div><div><div><br>Then the LKB generates everything subsumed by this:<br><div><div><div>He sleeps<br></div>She sleeps<br></div></div><div>They sleep<br><br></div><div>ACE generates only things with exactly the same semantics:<br>They sleep<br></div><div>(i actually generates the others then rejects them with the heart-warming message: <br>NICE TRY! `She sleeps.' has incompatible semantics:<br></div>NICE TRY! `He sleeps.' has incompatible semantics:<br><br></div><div>When translating from Japanese to English, we often have partial information about some property, and would like to specify to generate that and *get everything subsumed by it*. Not being able to do this is greatly reducing the coverage of JAEN using ACE: we often just set NUM to num, and this results in nothing being generated (as the number is always specified to something lower).<br><br></div><div>I am happy to also specifically submit this as an ACE feature request, but am trying to send things to developers by default.<br clear="all"></div><div><div><br></div><div>In the deep dark past Woodley said:<br>> The question of how to filter <span class="">generation</span> results in a useful way, though, is perhaps still an open one. Strictly checking for <span class="">subsumption</span>
by the input MRS is the right thing in most situations, I think. The
LKB has long implemented the option (which you presumably have been
using) to normally only output the subsumed subset, but output all the
results if none of them are subsumed. While a handy hack, this strikes
me as a unprincipled. Some kind of ranking of the outputs by how
severely they fail to be subsumed by the input would be better (sounds a
little like optimality theory).<br></div><div><br></div><div>It seems to me that currently ACE is checking for equality, not subsumption.<br></div><div><br><br></div><div>See also: <a href="http://moin.delph-in.net/SaarlandMrsSubsumptionDiscussion">http://moin.delph-in.net/SaarlandMrsSubsumptionDiscussion</a><br></div><div><br>-- <br><div class="gmail_signature">Francis Bond <<a href="http://www3.ntu.edu.sg/home/fcbond/">http://www3.ntu.edu.sg/home/fcbond/</a>><br>Division of Linguistics and Multilingual Studies<br>Nanyang Technological University</div>
</div></div></div></div></div></div>