Cascading versus Joint Modeling for Hierarchical Offensive Language Detection
Computation and Language
Summary
The authors studied two ways to detect different levels of offensive language: a step-by-step cascade method and a combined multi-task model. They carefully tested which method works better in terms of accuracy, model size, and speed. Their experiments showed the cascade method was more accurate, especially for the hardest task, but it required more computing power and was slower. They also found that picking how to handle uneven data just by guesswork isn’t the best approach and that their tested adjustments improved results. Overall, they showed there is a clear trade-off between better accuracy and higher computational cost.
Authors
Ruixing Ren, Junhui Zhao, Xiaoke Sun
Abstract
Fine-grained offensive language detection organizes labels into a hierarchical structure, for which two modeling paradigms exist: cascaded decomposition and joint multi-task modeling. Prior work rarely provides a direct, controlled comparison of the two paradigms in terms of accuracy, parameter count, and inference latency, and rarely verifies whether a chosen class-imbalance handling strategy is actually optimal. This paper proposes a three-level cascaded detection system whose training strategy is customized per subtask, together with two verification mechanisms. First, a controlled ablation study determines the best class-imbalance handling strategy for each subtask. Second, a joint multi-task model with a shared encoder is trained as an architectural control, yielding real measurements along the dimensions of accuracy, parameter count, and inference latency. Experiments show that the cascaded system attains macro-F1 scores of 0.795, 0.716, and 0.557 on the three subtasks of the official test set. The ablation study reveals that configuring the loss function purely by imbalance-severity intuition is suboptimal; reconfiguring based on the ablation results improves both performance and stability. End-to-end cascade evaluation shows that roughly one-fifth of the errors in the cascade pipeline originate from the first-stage filter and cannot be corrected by subsequent stages. Relative to the joint multi-task model, the cascaded architecture achieves higher accuracy on all three subtasks, with a 7.1-point macro-F1 gain on the most severely imbalanced subtask, at the cost of three times the parameters and 1.67 times the inference latency. Together, these results establish an explicit, quantifiable trade-off between the accuracy advantage of cascaded architectures and their deployment cost.