[rfc-i] Natural Language Processing (NLP) applied to RFCs
csp at csperkins.org
Tue Mar 26 23:56:21 PDT 2019
> On 26 Mar 2019, at 11:39, Michael Welzl <michawe at ifi.uio.no> wrote:
> I’m planning to do research with colleagues in NLP, with various ideas on applying it to RFCs (and, perhaps more importantly, Internet-drafts, which means that we can give feedback *before* RFC publication).
> One example outcome of this is that we could perhaps produce a tool that automatically gives feedback on language clarity - e.g. “You may want to check this sentence again, are you sure it specifies what should go into that header field clearly enough?” Heck, perhaps we can even give feedback like “shouldn’t this be an uppercase SHOULD here?”.
> I’d like to understand what people think about doing such stuff, in general - would a tool that gives some such feedback as a result of using NLP be useful to have?
> (obviously, *I* think so…)
I definitely think it could be useful. I’ve got a proposal under consideration, with Gareth Tyson and colleagues at QMUL, to try to get funding do something similar – we should sync-up.
More information about the rfc-interest