Figure 1From: Tackling racial bias in automated online hate detection: Towards fair and accurate detection of hateful users with geometric deep learningConfusion matrix. Predictions from best performing model: GraphSAGE with mean aggregationBack to article page