HierBias: Context-Conditioned Hierarchical Media Bias Detection with Multi-Task Type Classification

HierBias：基于上下文条件的分层媒体偏见检测与多任务类型分类

Abstract: Media bias detection is a critical task for ensuring fair and balanced information dissemination, yet existing sentence-level approaches classify each sentence independently, ignoring inter-sentence contextual signals that human annotators naturally exploit.

摘要： 媒体偏见检测是确保信息传播公平与平衡的关键任务。然而，现有的句子级方法通常独立地对每个句子进行分类，忽略了人类标注者在自然阅读时所利用的句子间上下文信号。

We present HierBias, a hierarchical context-conditioned media bias detector that formally models document context in bias prediction. We introduce the context-conditioned bias probability and prove theoretically that leveraging document context strictly reduces the Bayes error of sentence-level classification when inter-sentence mutual information is non-zero.

我们提出了 HierBias，这是一种分层上下文条件媒体偏见检测器，它在偏见预测中对文档上下文进行了形式化建模。我们引入了“上下文条件偏见概率”（context-conditioned bias probability），并从理论上证明，当句子间的互信息非零时，利用文档上下文可以严格降低句子级分类的贝叶斯误差。

A multi-task generalization bound further establishes that jointly training binary bias detection and fine-grained bias type classification improves sample efficiency on small annotated corpora.

多任务泛化界限进一步证明，联合训练二元偏见检测和细粒度偏见类型分类，可以提高在小型标注语料库上的样本效率。

Architecturally, HierBias pairs a sentence-level RoBERTa encoder with a cross-sentence Transformer aggregator and dual output heads for binary detection and four-class type classification.

在架构上，HierBias 将句子级的 RoBERTa 编码器与跨句 Transformer 聚合器相结合，并配备了用于二元检测和四分类任务的双输出头。

Evaluated on BABE and BASIL, HierBias achieves 0.853 F1 and 0.723 MCC, surpassing the state-of-the-art bias-detector by $+2.6%$ F1 and $+4.3%$ MCC (McNemar’s test, $p < 0.05$). Ablation experiments confirm that each theoretical component contributes independently and consistently.

在 BABE 和 BASIL 数据集上的评估显示，HierBias 达到了 0.853 的 F1 分数和 0.723 的 MCC，超过了当前最先进的偏见检测器（F1 提升 $+2.6%$，MCC 提升 $+4.3%$，McNemar 检验 $p < 0.05$）。消融实验证实，每个理论组件都独立且持续地做出了贡献。