Fig. 5

Feature importance of the XGBoost model. The vertical axis represents the value of the feature, ranging from low to high. The horizontal axis represents the feature’s impact on the model output. For example, a high value of “bert_322” (top-ranking feature, with the high value represented by red color) has a high negative impact across model predictions, with most SHAP values ranging between − 0.50 and − 1.00. The feature analysis shows the usefulness of BERT for online hate detection