Offensive Hebrew Corpus and Detection using BERT
This work addresses offensive language detection for Hebrew, a low-resource language, but is incremental as it applies existing BERT methods to new data.
The authors tackled the lack of offensive language detection resources in Hebrew by creating a new corpus of 15,881 labeled tweets and fine-tuning BERT models, achieving up to 69% accuracy and a 2% performance boost in some configurations.
Offensive language detection has been well studied in many languages, but it is lagging behind in low-resource languages, such as Hebrew. In this paper, we present a new offensive language corpus in Hebrew. A total of 15,881 tweets were retrieved from Twitter. Each was labeled with one or more of five classes (abusive, hate, violence, pornographic, or none offensive) by Arabic-Hebrew bilingual speakers. The annotation process was challenging as each annotator is expected to be familiar with the Israeli culture, politics, and practices to understand the context of each tweet. We fine-tuned two Hebrew BERT models, HeBERT and AlephBERT, using our proposed dataset and another published dataset. We observed that our data boosts HeBERT performance by 2% when combined with D_OLaH. Fine-tuning AlephBERT on our data and testing on D_OLaH yields 69% accuracy, while fine-tuning on D_OLaH and testing on our data yields 57% accuracy, which may be an indication to the generalizability our data offers. Our dataset and fine-tuned models are available on GitHub and Huggingface.