UAntwerp at SemEval-2021 Task 5: Spans are Spans, stacking a binary word level approach to toxic span detection

Ben Burtenshaw, Mike Kestemont


Abstract
This paper describes the system developed by the Antwerp Centre for Digital humanities and literary Criticism [UAntwerp] for toxic span detection. We used a stacked generalisation ensemble of five component models, with two distinct interpretations of the task. Two models attempted to predict binary word toxicity based on ngram sequences, whilst 3 categorical span based models were trained to predict toxic token labels based on complete sequence tokens. The five models’ predictions were ensembled within an LSTM model. As well as describing the system, we perform error analysis to explore model performance in relation to textual features. The system described in this paper scored 0.6755 and ranked 26th.
Anthology ID:
2021.semeval-1.121
Volume:
Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021)
Month:
August
Year:
2021
Address:
Online
Venues:
ACL | IJCNLP | SemEval
SIG:
SIGLEX
Publisher:
Association for Computational Linguistics
Note:
Pages:
898–903
Language:
URL:
https://aclanthology.org/2021.semeval-1.121
DOI:
10.18653/v1/2021.semeval-1.121
Bibkey:
Copy Citation:
PDF:
https://aclanthology.org/2021.semeval-1.121.pdf