Re: Fwd: Legal advice regarding the NEW queue
Maybe it is doable. In my impression the problem of mapping a tree
of files into a serialized attribute report sounds like the
part-of-speech tagging problem in computational linguistics.
Combining the modeling with the state-of-the-art language transformers
with the debian archive being the training dataset should make
sense.
I noticed that the post is off from -devel,
so I felt free do say so.
On Fri, 2022-02-04 at 01:02 +0100, Maxime Chambonnet wrote:
>
> I can't really address that, but there are other reasons than
> legal
> risks to follow Debian policy. My impression is that people are
> tired of waiting on New, but no one really seems to be interested
> in
> doing any work on any alternative other than more bugs. I'm not
> sure
> how a lawyer can tell you how many bugs are acceptable.
>
> Scott K
>
> The path forward with this topic is maybe to have a neural network
> inference that writes d/copyright with 99.9% accuracy! 😉
>
Reply to: