The Cross-Domain Generalization Cost of Offensive Language Detection

Offensive language detection models generally suffer performance degradation when deployed across datasets and across languages, yet most existing studies stop at reporting this phenomenon and lack a systematic methodology for decomposing the causes of degradation into attributable components and quantifying the cost of remediation. This paper proposes a diagnosis and optimization framework composed of three coordinated technical components. First, a zero-shot transfer loss decomposition that separates the performance degradation from OLID to MLMA into two independently measurable components, namely dataset effect and language effect. Second, a controlled fine-tuning protocol that quantifies both adaptation efficiency and the hidden damage inflicted on the source task by comparing few shot learning curves under continued fine-tuning and cold-start starting points. Third, three joint training strategies incorpo rating temperature sampling and experience replay, which offer a controllable Pareto trade-off between improving multilingual capability and preserving source-task performance. Experiments built on this framework show that the dataset effect dominates the zero-shot transfer loss and substantially outweighs the language effect. Few-shot adaptation without a replay mechanism, though data-efficient, inflicts source task damage 4 to 9 times greater than that of the joint training strategies, and its damage magnitude is highly unstable. The three joint training strategies trade 3.2 to 4.1 percentage points of source-task performance for 8.1 to 42.6 percentage points of multilingual capability gain, forming a clear and controllable Pareto trade-off.

Cascading versus Joint Modeling for Hierarchical Offensive Language Detection

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.

NLP-Based Review for Toxic Comment Detection Tailored to the Chinese Cyberspace

With the in-depth integration of mobile Internet and widespread adoption of social platforms, user-generated content in the Chinese cyberspace has witnessed explosive growth. Among this content, the proliferation of toxic comments poses severe challenges to individual mental health, community atmosphere and social trust. Owing to the strong context dependence, cultural specificity and rapid evolution of Chinese cyber language, toxic expressions are often conveyed through complex forms such as homophones and metaphors, imposing notable limitations on traditional detection methods. To address this issue, this review focuses on the core topic of natural language processing based toxic comment detection in the Chinese cyberspace, systematically collating and critically analyzing the research progress and key challenges in this field. This review first defines the connotation and characteristics of Chinese toxic comments, and analyzes the platform ecology and transmission mechanisms they rely on. It then comprehensively reviews the construction methods and limitations of existing public datasets, and proposes a novel fine-grained and scalable framework for toxic comment definition and classification, along with corresponding data annotation and quality assessment strategies. We systematically summarize the evolutionary path of detection models from traditional methods to deep learning, with special emphasis on the importance of interpretability in model design. Finally, we thoroughly discuss the open challenges faced by current research and provide forward-looking suggestions for future research directions.