On the Evaluation of Common-Sense Reasoning in Natural Language Understanding

29 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

WinoGrande (2019, methodology): “) This is the original Winograd Schema Challenge dataset, which consists of 273 problems. The problems are manually crafted by the authors to avoid word association bias as much as possible, although Trichelair et al. (2018) later report that 13.5% of the questions may still have word-association bi”
An Adversarial Winograd Schema Challenge at Scale (2019, methodology): “Language-based bias, or word association bias (Trichelair et al. (2018)), refers to the case where the correct answer aligns with more frequent patterns in natural language, thus can be easily solved by neural language models trained over large corpora (Table 1 (3) and (4)).”
The KnowRef Coreference Corpus: Removing Gender and Number Cues for Difficult Pronominal Anaphora Resolution (2018, methodology): “In the case of pronominal anaphora resolution, these forms of “common sense” become much more * equal contribution important when cues like gender and number do not by themselves indicate the correct resolution (Trichelair et al., 2018).”
Language Models are Unsupervised Multitask Learners (2019, methodology): “The dataset is quite small with only 273 examples so we recommend reading Trichelair et al. (2018) to help contextualize this result.”
Contrastive Self-Supervised Learning for Commonsense Reasoning (2020, background): “However, according to (Trichelair et al., 2018), it is less challenging due to several inherent biases.”