TY - JOUR
AU - 
AB - In this paper, we present the first compre- hensive categorization of essential common- sense knowledge for answering the Winograd Schema Challenge (WSC). For each of the Figure 1: A pair of questions in WSC. questions, we invite annotators to first pro- vide reasons for making correct decisions and then categorize them into six major knowl- 2019). Among all developed commonsense rea- edge categories. By doing so, we better un- derstand the limitation of existing methods soning tasks, the Winograd Schema Challenge (i.e., what kind of knowledge cannot be ef- (WSC) (Levesque et al., 2012), which is a hard fectively represented or inferred with existing pronoun coreference resolution task, is one of the methods) and shed some light on the com- most influential ones. All questions in WSC are monsense knowledge that we need to acquire grouped into pairs such that paired questions have in the future for better commonsense reason- minor differences (mostly one-word difference), ing. Moreover, to investigate whether cur- but reversed answers. For each question, we de- rent WSC models can understand the com- monsense or they simply solve the WSC ques- note the other question in the same pair as its tions based on the statistical 
TI - WinoWhy: A Deep Diagnosis of Essential Commonsense Knowledge for Answering Winograd Schema Challenge
JF - Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics
DO - 10.18653/v1/2020.acl-main.508
DA - 2020-01-01
UR - https://www.deepdyve.com/lp/unpaywall/winowhy-a-deep-diagnosis-of-essential-commonsense-knowledge-for-0AlN0Gmwmb
DP - DeepDyve
ER -