On 11 July 2010 18:18, Jimmy O'Regan <[email protected]> wrote: > Motivation: > > The primary motivation was in dealing with Polish: highly inflected > (few 'markers'), adjectives can come before or after the noun. > Inflection *usually* gives enough information for proper segmentation, > but handling it properly would be a matter of having individual rules > for each gender, case, and number + each combination of words (i.e., > multiply number of NP rules by 70). I've seen recently that it would > help in less inflected languages, so it's probably generally useful. >
Ok, that sucked as an explanation. Let's take a few samples: matka młodego chłopaka matka[n.f.sg.nom] młody[adj.mp.sg.gen] chłopak[n.mp.sg.gen] ([the] mother of that boy) Using a rule that matches adj+n will segment that badly: (matka młodego) (chłopaka) Having an agreement check inside a normal rule could give (matka) (młodego) (chłopaka) but adds unneeded work in t2 to rejoin (młodego chłopaka) An agreement check in <exception> can solve this problem. Where it won't work, however, is for something like: dla ojca młodego chłopaka dla[prep.gen] ojciec[n.mp.sg.gen] młody[adj.mp.sg.gen] chłopak[n.mp.sg.gen] (for [the] father of that boy) ...but this is the sort of thing that doesn't affect most languages, and requires much ugliness to work around no matter what else works. > Caveats: > > * Sanity checks aren't -- more are needed > * Only in apertium-transfer (I'll add this to interchunk when transfer > is tested) > * Only a single level of backoff (if the rule that's being backed off > to also triggers an exception... breakage will happen). The plan here is to change the two extra variables I added into a stack of pairs, but I wanted something that gives an idea of how it will work before I fix it up properly. -- <Leftmost> jimregan, that's because deep inside you, you are evil. <Leftmost> Also not-so-deep inside you. ------------------------------------------------------------------------------ This SF.net email is sponsored by Sprint What will you do first with EVO, the first 4G phone? Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first _______________________________________________ Apertium-stuff mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/apertium-stuff
