On 2 September 2013 15:49, Xavi Ivars <xavi.iv...@gmail.com> wrote: > 2013/9/2 Jimmy O'Regan <jore...@gmail.com> >> >> >> > The first test >> > sentences for the Block World Corpus are better translated by the >> > outdated statistical translation model IBM model 1 in the direction >> > French to Spanish. Apparently, Apertium has some problems with the >> > imperative of verbs and goes for the subjunctive used in negated >> > requests (this problem persists in the omitted sentences): >> > >> > Original: >> > >> > prenez une flèche >> > prenez un bloc >> > prenez un cône bleu >> > >> >> Your sentences are not terminated. If they had been, you would have >> seen the output you expected. >> >> > > This is something I noticed when the first email was sent, but I didn't look > deeper in it: it doesn't make sense than in a rule-based engine like > Apertium "prenez" was translated sometimes as "tomad" and sometimes as > "tomáis", when in fact one of the (key?) benefits of the rule-based systems > is predictability on the output.
It's not hard to understand. There are no sentence boundaries, making this one big sentence. The tagger sees something that may be an imperative following a noun and, because that's unlikely, chooses something else instead (as would a rule-based disambiguator with a reasonable set of rules). Garbage in, garbage out. -- <Sefam> Are any of the mentors around? <jimregan> yes, they're the ones trolling you ------------------------------------------------------------------------------ Learn the latest--Visual Studio 2012, SharePoint 2013, SQL 2012, more! Discover the easy way to master current and previous Microsoft technologies and advance your career. Get an incredible 1,500+ hours of step-by-step tutorial videos with LearnDevNow. Subscribe today and save! http://pubads.g.doubleclick.net/gampad/clk?id=58040911&iu=/4140/ostg.clktrk _______________________________________________ Apertium-stuff mailing list Apertium-stuff@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/apertium-stuff
