On Sun, Feb 02, 2020 at 09:46:19PM -0500, Daniel Swanson wrote:

> I was recently experimenting with alternative ways of compiling analyzers
> for languages with prefixes and I've come up with something that produces
> results equivalent to using path restriction rules but compiles
> substantially faster (up to 600x for Navajo and Lingala).

That seems interesting, I was thinking of a similar thing back when we
were "reverse engineering" lexc, but actually never had the time to
implement it. I haven't tested it much beyond the
included tests but it seems promising. Do you have plans on doing tests
on runtime efficiency, i.e. how fast it is to run the automata on texts?
One thing that we found with flag diacritics on lexc is is that it's
kindof possible to abuse them to optimise the compiled stuff and it'd
probably be interesting to see here too, I see there's something with
flags in the code already?


> I'm interested in any feedback on the syntax or anything else that might
> make this more useful to people.

Well, I have two self-contradicting opinions of lexc-like syntaxes versus
XMLs. On the other hand I think most of XML usages just look unappealing
and wrong, but if I need to teach the stuff, having XML for all of the
components is probably easier than collection of slightly
differing text formats. But anyways, the format seems good and clean,
and I like it how things end up looking like linguistic explanation of
the morphotactics with named patterns and all. 

Cool stuff, it'd be interesting to see something like this in standard
language packages
-- 
Doktor Tommi A Pirinen, Computational Linguist,
<https://flammie.github.io/purplemonkeydishwasher/>, Universität
Hamburg, Hamburger Zentrum für Sprachkorpora <http://hzsk.de>. CLARIN-D
Entwickler.  President of ACL SIGUR SIG for Uralic languages
<http://gtweb.uit.no/sigur/>.
I tend to follow inline-posting style in desktop e-mail messages.

Attachment: signature.asc
Description: PGP signature

_______________________________________________
Apertium-stuff mailing list
Apertium-stuff@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/apertium-stuff

Reply via email to