On Sat, Sep 11, 2010 at 7:12 PM, Kevin Reid
<kpreid@switchb.org> wrote:
On Sep 11, 2010, at 21:02, Robin Lee Powell wrote:
I want to be able to generate *at least* a formal dictionary,
flashcards, simple definitions, and glossing, from one datasource,
easily, with the data source being as simple as possible.
Aaaaaaah. I see now. Such as how smart.fm needed additional text to describe each place of a gismu.
I think that this problem falls under NLP, because the text needed in each case is not *regular* (that is, related in a systematic way to the other cases). For example, I've tweaked the smart.fm definitions because the existing more-regular (though still hand-written) ones were misleading in English. (I don't think I kept notes on which ones I changed; the only example I remember was x2 of troci => "something tried", which is wrong but not a good example of the generic type of wrongness.)
Well, there is a reason why I made the goal open access..