Hi, On Thu, May 30, 2013, at 19:47, Francis Tyers wrote: > El dj 30 de 05 de 2013 a les 19:42 +0200, en/na Per Tunedal va escriure: > > Hi, > > wouldn't it be possible to create rules for the translation of names, > > somehow? Numbers are nicely treated in the Swedish-Danish (sv-da) pair, > > but not proper names. It's kind of crazy to try to enter all kinds of > > names any Dane or Swede might have. > > > > I would like to have some clever rule that just passes them over. I'm > > Per in Danish as well as in Swedish! > > (OK, some geographical names might need to be translated, they can be > > put in the dictionaries, just as an exception.) > > > > Probably it looks just about the same in other language pairs, except > > for that names sometimes have to be translated or transliterated. It > > could be done according to some rule as well, I suppose. > > > > For the exceptions: I suppose the rule could be skipped if there is an > > entry in the dictionary. > > > > The most difficult part would be to find the names. Perhaps someone has > > any ideas? > > In Icelandic--English, regular expressions are used. See e.g. pardefs > for "persons" and "lastnames" in is.dix > > This is not altogether recommended though, as regular expressions slow > down your transducer. What you could do is use them on a large corpus > and then mass-add the ones after superficial checking. > > Fran
I'm reading Philipp Koehn's book on Statistical Machine Translation, he suggests a finite state transducer for transliteration. Wouldn't that fit nicely into Apertium? Yours, Per Tunedal ------------------------------------------------------------------------------ Introducing AppDynamics Lite, a free troubleshooting tool for Java/.NET Get 100% visibility into your production application - at no cost. Code-level diagnostics for performance bottlenecks with <2% overhead Download for free and get started troubleshooting in minutes. http://p.sf.net/sfu/appdyn_d2d_ap1 _______________________________________________ Apertium-stuff mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/apertium-stuff
