Missing Information, Unresponsive Authors, Experimental Flaws: The Impossibility of Assessing the Reproducibility of Previous Human Evaluations in NLP
Belz, Anya; Thomson, Craig; Reiter, Ehud; Abercrombie, Gavin; Alonso-Moral, Jose M.; Arvan, Mohammad; Cheung, Jackie; Cieliebak, Mark; Clark, Elizabeth; Deemter, Kees van; Dinkar, Tanvi; Dušek, Ondřej; Eger, Steffen; Fang, Qixiang; Gatt, Albert; Gkatzia, Dimitra; González-Corbelle, Javier; Hovy, Dirk; Hürlimann, Manuela; Ito, Takumi; Kelleher, John D.; Klubicka, Filip; Lai, Huiyuan; Lee, Chris van der; Miltenburg, Emiel van; Li, Yiru; Mahamood, Saad; Mieskes, Margot; Nissim, Malvina; Parde, Natalie; Plátek, Ondřej; Rieser, Verena; Romero, Pablo Mosteiro; Tetreault, Joel; Toral, Antonio; Wan, Xiaojun; Wanner, Leo; Watson, Lewis; Yang, Diyi
(2023) The Fourth Workshop on Insights from Negative Results in NLP, pp. 1 - 10
(Part of book)
Abstract
We report our efforts in identifying a set of previous human evaluations in NLP that would be suitable for a coordinated study examining what makes human evaluations in NLP more/less reproducible. We present our results and findings, which include that just 13% of papers had (i) sufficiently low barriers to
... read more
reproduction, and (ii) enough obtainable information, to be considered for reproduction, and that all but one of the experiments we selected for reproduction was discovered to have flaws that made the meaningfulness of conducting a reproduction questionable. As a result, we had to change our coordinated study design from a reproduce approach to a standardise-then-reproduce-twice approach. Our overall (negative) finding that the great majority of human evaluations in NLP is not repeatable and/or not reproducible and/or too flawed to justify reproduction, paints a dire picture, but presents an opportunity for a rethink about how to design and report human evaluations in NLP.
show less
Download/Full Text
Publisher: Association for Computational Linguistics
(Peer reviewed)