On 03/08/2012 11:48 PM, Gary Gregory wrote: [snip]
> Would you be willing to handle this merge and describing the algorithm in > the Nysiis class itself. The description in the test method feels out of > place. > > After that, I think I'll put a message out on the ML and ask for further > testing and feedback. yes, I will do it, tomorrow ;-) >> I have not found the original paper, which is a pity, and all the >> algorithm descriptions I have found so far vary a bit. But in the end, >> it's a phonetic code to match similar names and when I compare to >> dropby, I feel more comfortable with our implementation (e.g. take PHIL >> and FIL which result in FFAL and FAL in dropby, which is weird). >> >> Anyway, the modified version seems to address some of these things, so >> it may be a good idea to additionally implement this one. >> > > What do you mean? > > IMO, we should have one impl that is documented. If it deviates from the > 'standard', then we should document that. that's fine for me, I just mentioned the modified variant, as Henri was referring to it in a comment to the issue. > This is why I added the 'trueLength' (lame name?) ivar because that seems > like a reasonable toggle after reading the Wikipedia entry. I wanted to add this myself, but you were quicker ;-) Good night, Thomas --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
