Hello, I am looking at the Xapian database spelling support, with the idea of getting rid of the cumbersome use of aspell in my application. One advantage of aspell, though, (in my eye), is that it can supply a list of alternate spellings, not just the one it considers best. In some cases, this can be useful. 2 examples on my test index: porbability: xapian suggests portability, which is nice, but I was intending probability which is in the aspell list all right. macdonell: xapian suggests macdonald, all right again, but aspell also suggests mcdowell and mcdonald, which might be actually useful to someone trying to remember the name of a correspondant. When just trying to quick-fix the query, or running an alternate query like Google does, using the "best" candidate is fine. But having another mode to let the user choose among a list of possibles is nice too. Any chance that Xapian would support this ? Cheers, jf
On Mon, May 09, 2011 at 08:08:49AM +0200, Jean-Francois Dockes wrote:> When just trying to quick-fix the query, or running an alternate query like > Google does, using the "best" candidate is fine. But having another mode to > let the user choose among a list of possibles is nice too. > > Any chance that Xapian would support this ?Yes - this is one of the things which Nikita is planning to address in GSoC: http://www.google-melange.com/gsoc/project/google/gsoc2011/nsmetanin/16001> In some cases, this can be useful. 2 examples on my test index: > > porbability: xapian suggests portability, which is nice, but I was > intending probability which is in the aspell list all right.I think in this case, the transposition is a more likely mistake than typing 'b' rather than 't', which is something else which Nikita is planning to look at.> macdonell: xapian suggests macdonald, all right again, but aspell also > suggests mcdowell and mcdonald, which might be actually useful to someone > trying to remember the name of a correspondant.Phonetic similarity was another one. Cheers, Olly