Re: [Openadaptxt-linguists] Phrases & text

[Michael B. <fi...@ak...>]

Makes sense, thanks!

Michael

08/02/2012 09:37, sgrìobh Jens Christensen:
> Hi Michael,
> Yes, you will get strange stuff like that, even if you set the cutoff quite high. Of course the higher the cutoff the fewer you should get, but then you will also get less of the "good" ones, so it's a trade-off either way. If you want to remove the oddities I can't really think of any other way of doing that than reviewing it manually. It's up to yourselves to judge what would be necessary to remove, if any, and whether it's necessary to review the context - you are the experts:-)
>
> As you noted, the current dictionaries (not just the English one) only contain a very small amount of context, basically the very most common. This is both to avoid "oddities" as you mention but more to leave the dictionaries open for people like yourselves or anybody else who might be interested in using or working with openadaptxt. Basically we didn't want to say "this is the final dictionaries" but rather leave it to the community what to do with the dictionaries.
>
> Cheers,
> Jens