On 05/21/2013 04:55 PM, Kevin Atkinson wrote:
Aspell does not support what I think you are after very well. That is why it is not enabled by default in the English dictionary. For some insight on why see: http://aspell.net/man-html/Words-With-Symbols-in-Them.html
I think that you started out with the goal of making a spell checker, and for that you need tokenization, and so you need to distinguish letters and space and punctuation. But for grammar checking or translation, you need to be able to find phrases and patterns, not just tokens. Perhaps the basic layer of a larger, more generic natural language processing library should have a matching algorithm that isn't based on tokenization. I have no idea if such libraries exist or what the current state of the art is. But I think it's natural for a dictionary to contain both words (cat, dog) and phrases (kill two birds with one stone), so that "kill two dogs with one stone" will generate a warning that perhaps the user meant to write "birds" there. That would certainly be another type of application than Aspell, but perhaps a more useful one. -- Lars Aronsson (l...@aronsson.se) Aronsson Datateknik - http://aronsson.se _______________________________________________ Aspell-user mailing list Aspell-user@gnu.org https://lists.gnu.org/mailman/listinfo/aspell-user