There is a newer version of this record available.

Conference paper Open Access

Bayesian BERT for Trustful Hate Speech Detection

Miok, Kristian; Škrlj, Blaž; Zaharie, Daniela; Robnik-Šikonja, Marko

Hate speech is an important problem in the management of user-generated content. In order to remove offensive content or ban misbehaving users, content moderators need reliable hate speech detectors. Recently, deep neural networks based on transformer architecture, such as (multilingual) BERT model, achieve superior performance in many natural language classification tasks, including hate speech detection. So far, these methods have not been able to quantify their output in terms of reliability. We propose a Bayesian method using Monte Carlo Dropout within the attention layers of the transformer models to provide well-calibrated reliability estimates. We evaluate the introduced approach on hate speech detection problems in several languages. Our approach not only improves the classification performance of the state-of-the-art multilingual BERT model but the computed reliability scores also significantly reduce the workload in inspection of offending cases and in reannotation campaigns.

Files (581.3 kB)
Name Size
Miok.pdf
md5:5e8e533a690f9a2078dc4c59b6c7eceb
581.3 kB Download
60
45
views
downloads
All versions This version
Views 6013
Downloads 4514
Data volume 26.5 MB8.1 MB
Unique views 479
Unique downloads 3914

Share

Cite as