W dniu 08.03.2016 o 17:43, Daniel Naber pisze: > On 2016-03-08 00:11, Jaume Ortolà i Font wrote: > > Hi Jaume, > >> The Catalan dictionary is 1.1M with CFSA2 and 1.4M with FSA5. What >> should we use? I don't know if the "the cost of traversing" the >> dictionary is relevant. > > I'd suggest to use the smaller one. We won't need to re-build all the > dictionaries, do we? You could use PerformanceCheck.java to see if > there's a difference in performance, although you'd probably need to run > it several times due to random variation.
I think it's almost completely irrelevant. And for some languages, the differences are much bigger (e.g., for Polish), so fsa5 is definitely not the best format. So please go ahead with CFSA2. Best, Marcin ------------------------------------------------------------------------------ Transform Data into Opportunity. Accelerate data analysis in your applications with Intel Data Analytics Acceleration Library. Click to learn more. http://makebettercode.com/inteldaal-eval _______________________________________________ Languagetool-devel mailing list Languagetool-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/languagetool-devel
