HLE-UPC at SemEval-2021 Task 5: Multi-Depth DistilBERT for Toxic Spans Detection

Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021), pages 960–966 Bangkok, Thailand (online), August 5–6, 2021. ©2021 Association for Computational Linguistics 960 HLE-UPC at SemEval-2021 Task 5: Multi-Depth DistilBERT for Toxic Spans Detection Rafel Palliser-Sans and Albert Rial-Farr`as Facultat d’Inform`atica de Barcelona (FIB) Universitat Polit`ecnica de Catalunya (UPC - BarcelonaTECH) Barcelona, Spain {rafel.palliser, albert.rial}@estudiantat.upc.edu Abstract This paper presents our submission to SemEval-2021 Task 5: Toxic Spans Detection. The purpose of this task is to detect the spans that make a text toxic, which is a complex labour for several reasons. Firstly, because of the intrinsic subjectivity of toxicity, and secondly, due to toxicity not always coming from single words like insults or offends, but sometimes from whole expressions formed by words that may not be toxic individually. Following this idea of focusing on both single words and multi-word expressions, we study the impact of using a multi-depth DistilBERT model, which uses embeddings from different layers to estimate the final per-token toxicity. Our quantitative results show that using information from multiple depths boosts the performance of the model. Finally, we also analyze our best model qualitatively. 1 Introduction SemEval-2021 Task 5: Toxic Spans Detection (Pavlopoulos et al., 2021) consists in detecting which spans make a text toxic. This is quite rele- vant for nowadays lifestyle in which, aggravated by the COVID-19 pandemic, online conversations have become key to communicate with our family, friends and job mates, or socialize through social networks and streaming chats. Being able to mod- erate all this digital content is crucial in order to promote healthy online conversations and discus- sions. To tackle this problem, in HLE-UPC we have used a BERT-based model with a fully-connected layer on top to perform Named-Entity Recognition and Classification (NERC), with the goal of tag- ging each word as either toxic or not. Moreover, we have studied and proved that the use of informa- tion from different-depth layers enriches the final classification. Our contributions to Toxic Spans Detection are: • The proposal of an ensemble of three different multi-depth DistilBERTs, achieving an F1- score of 68.54% and being ranked 14th out of 91 teams in the challenge, just 2.29% below the best performing model. • The study of multi-depth BERT-based models in the task of Toxic Spans Detection, show- ing an improvement on the performance com- pared to non-multi-depth architectures. • A qualitative analysis presenting some ethical concerns regarding racial bias. The source code for our model and pipeline is available at https://github.com/rafelps/ HLE-UPC-SemEval-2021-ToxicSpansDetection. 2 Related work Toxicity The task in which we are participating is not the first one to focus on text toxicity. Without going any farther, in last year’s edition of SemEval we can find Task 12, also known as OffensEval 2020 (Zampieri et al., 2020), in which the goal was to identify offensive language in multilingual social media data. In the previous year’s competition, SemEval 2019, Task 6 (Zampieri et al., 2019) was also tackling the identification and categorization of offensive language in social media. …

Content truncated to 500 words