El dg 30 de 09 de 2012 a les 08:40 +0200, en/na Mikel Forcada va escriure: > Al 09/30/2012 01:13 AM, En/na Francis Tyers ha escrit:
> > Morph: > > > > ^si/si<cnjadv>$ > > ^lo/lo<detnt>/lo<prn><pro><p3><nt>/lo<prn><pro><p3><m><sg>$ > > ^fuera/fuera<adv>/ir<vblex><pis><p1><sg>/ir<vblex><pis><p3><sg>/ser<vbser><pis><p3><sg>/ser<vbser><pis><p1><sg>$^./.<sent>$ > > > > Tagger: > > > > ^si<cnjadv>$ ^lo<prn><pro><p3><nt>$ ^ir<vblex><pis><p3><sg>$^.<sent>$ > > > > Hmm, we can't really forbid "lo + ir<vblex>" because of "lo vas a ver". > And many others. It is quite common as 'ir a' is used for informal future. Yes... > > But I can't help but think that this could be taken care of in a > > rule-based manner. > I cannot think of a good general rule based solution. I have discarded a > multiword for "Si lo fuera" which is quite common, but it would not work > If one had "si lo fuera a ver mañana" → "si l'anara a veure demà" OK. I think that if we have the sequence: si lo fuera , +cni Then we could discard the "ir" reading. We have a delimited clause (between the 'si' and the ", +cni") and in between that we have two possibilities, a verb which can take "ho" and one that can't -- in the examples "ho" goes with the main verb, not with 'ir'. Can you come up with any exceptions ? The problem is that this would take into account more context than we have for the tagger. Some comparison with other systems: Lucy: Si l'anés, hauríem de considerar... Goog: Si ho fos, hauríem de considerar ... Bing: Si fos així, hauria de considerar... > I would however add multiwords for fossilized expressions such as "Si > así fuese" or "Si así fuera". > > Sorry I cannot help much No problem, it's good to have a record of these things in any case one day we can treat them! :) F. ------------------------------------------------------------------------------ Everyone hates slow websites. So do we. Make your web apps faster with AppDynamics Download AppDynamics Lite for free today: http://ad.doubleclick.net/clk;258768047;13503038;j? http://info.appdynamics.com/FreeJavaPerformanceDownload.html _______________________________________________ Apertium-stuff mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/apertium-stuff
