Maintaining quality in FEVER annotation
Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review
Standard
Maintaining quality in FEVER annotation. / Schulte, Henri; Binau, Julie; Derczynski, Leon.
Proceedings of the Third Workshop on Fact Extraction and VERification (FEVER): Association for Computational Linguistics. Association for Computational Linguistics, 2020. p. 42-46.Research output: Conference Article in Proceeding or Book/Report chapter › Article in proceedings › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Maintaining quality in FEVER annotation
AU - Schulte, Henri
AU - Binau, Julie
AU - Derczynski, Leon
PY - 2020/7/9
Y1 - 2020/7/9
N2 - We propose two measures for measuring the quality of constructed claims in the FEVER task. Annotating data for this task involves the creation of supporting and refuting claims over a set of evidence. Automatic annotation processes often leave superficial patterns in data, which learning systems can detect instead of performing the underlying task. Humans also can leave these superficial patterns, either voluntarily or involuntarily (due to e.g. fatigue). The two measures introduced attempt to detect the impact of these superficial patterns. One is a new information-theoretic and distributionality based measure, DCI; and the other an extension of neural probing work over the ARCT task, utility. We demonstrate these measures over a recent major dataset, that from the English FEVER task in 2019.
AB - We propose two measures for measuring the quality of constructed claims in the FEVER task. Annotating data for this task involves the creation of supporting and refuting claims over a set of evidence. Automatic annotation processes often leave superficial patterns in data, which learning systems can detect instead of performing the underlying task. Humans also can leave these superficial patterns, either voluntarily or involuntarily (due to e.g. fatigue). The two measures introduced attempt to detect the impact of these superficial patterns. One is a new information-theoretic and distributionality based measure, DCI; and the other an extension of neural probing work over the ARCT task, utility. We demonstrate these measures over a recent major dataset, that from the English FEVER task in 2019.
M3 - Article in proceedings
SP - 42
EP - 46
BT - Proceedings of the Third Workshop on Fact Extraction and VERification (FEVER)
PB - Association for Computational Linguistics
ER -
ID: 85287454