+ new component
Today at LREC, Tanvi Dinkar presented our counterspeech paper "Can NLP Tackle Hate Speech in the Real World?" www.lrec-conf.org/proceedings/...
We have arrived at the Palau de Congressos for NLPerspectives 5, which starts in less than an hour!
Thanks for coming to my poster! (8/7)
There's a long list of side effects to this one (7/7)
The paper looks at the current state of 'perspectivism' after 5 editions of the workshop (2/7)
On Tuesday I had a poster at @nlperspectives.bsky.social workshop. Paper is available at lrec-conf.org/proceedings/...
I argue that we're often falling between the old 'gold label' data paradigm and rigorous analysis of the populations we're trying to model (3/7)
Not everything that people disagree on should be called subjective. See also aclanthology.org/2024.woah-1.... (6/7)
We're evaluating in ways that don't easily transfer to real-world applications (5/7)
Having accepted that many or even all annotator responses may be valid, we have new challenges surrounding how to detect noisy annotations (4/7)
Gavin Abercrombie
Gavin Abercrombie
Gavin Abercrombie
Gavin Abercrombie
Gavin Abercrombie
Gavin Abercrombie
Gavin Abercrombie
Gavin Abercrombie
Gavin Abercrombie