Open kiragrif opened 9 years ago
I would suggest to substitute the intended profanity as much as possible.
Mark "concepts" that are low on meaning and high on emotion/profanity with :mode expressive
.
Make sure to capture the actual meaning (e.g. and ****
is sometimes used a vulgar form of et-cetera
).
making up ****
(m / make-up-10
:ARG1 (s / shit :mode expressive))
That's ********
(f / fuck-up-02
:ARG1 (t / that))
aspirations and ****
(a / and
:op1 (t / thing
:ARG1-of (a2 / aspire-01))
:op2 (e / et-cetera))
I've been wearing the same ****ing clothes.
(w / wear-01
:ARG0 (i / i)
:ARG1 (c / clothes
:ARG1-of (s / same-01)
:mod (f / fucking :mode expressive)))
When the intended word is unclear, I suggest to use the concept 'expletive':
That's ********
(e / expletive
:domain (t / that))
At the AMR meeting on Sept. 28, 2015, we agreed to expand/"spell-correct" from **\ to intended meaning if possible.
No consensus yet on ":mode expressive" yet.
Hi all, wondering how we should annotate "obscenicons" (use of typographical symbols to represent profanities)?
There is a small amount of mixed precedent for this, with some annotators using the (likely) intended profanity, some substituting (context appropriate) non-profanity, and annotating the obscenicon string directly.
Substituting the intended profanity:
Substituting a non-profanity:
Annotating the string directly:
Determining the intended profanity might not always be straightforward, maybe we'd want a special concept or role for these. Thoughts?