arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2606.03576cs.CL
收藏

AutoTail-BSFGM:面向中文学术文本分类的类别平衡感知微调

AutoTail-BSFGM: Class-Balance-Aware Fine-Tuning for Chinese Scholarly Text Classification

Department of Intelligent Communication, School of Journalism and Communication, Minzu University of China(中国民族大学新闻与传播学院智能通信系);ZeeLin (Beijing) Technology Co., Ltd.(北京智联科技有限公司);School of Journalism and Communication, Tsinghua University(清华大学新闻与传播学院);College of AI, Tsinghua University(清华大学人工智能学院)

Anling Xiang, Yuwen Yang, Yang Shen

AI总结:

提出AutoTail-BSFGM方法,通过自动门控尾部调整、弱平衡Softmax辅助损失和快速梯度法对抗正则化,解决中文学术文本分类中的类别不平衡和语义邻近问题,在CSL数据集上提升了验证集和锁箱集准确率。

摘要:

学术文本分类支持文献组织、主题标引和研究情报,但中文学术语料库通常包含不平衡且语义邻近的学科标签。我们提出AutoTail-BSFGM,一种类别平衡感知的微调方法,它结合了自动门控尾部先验调整、弱平衡Softmax辅助损失和快速梯度法对抗正则化。该方法仅改变训练目标和过程;推理使用与相应标签平滑基线相同的单一基础规模编码器和线性分类器。我们在两个基于CSL的任务上评估该方法:一个包含67个标签的摘要到学科任务和一个包含13个类别的标题到类别任务。在主要的摘要任务上,AutoTail-BSFGM在中文RoBERTa-WWM和MacBERT-base下均提高了验证集和锁箱集准确率。使用MacBERT-base时,验证集准确率提高0.83个百分点,锁箱集准确率提高0.49个百分点,验证集上的合并配对McNemar检验显著(p = 0.023)。在标题任务上,该方法将验证集准确率提高0.70个百分点,验证集平衡准确率提高2.64个百分点;锁箱集准确率大致中性,而锁箱集平衡准确率提高1.22个百分点。结果支持有界贡献:AutoTail-BSFGM改善了类别平衡敏感行为,并在基于摘要的学术分类中取得一致增益,但并非在每个划分上均匀改善每个指标。

英文摘要:

Scholarly text classification supports literature organization, subject indexing, and research intelligence, but Chinese scholarly corpora often contain imbalanced and semantically adjacent disciplinary labels. We propose AutoTail-BSFGM, a class-balance-aware fine-tuning method that combines an automatically gated tail-prior adjustment, a weak Balanced Softmax auxiliary loss, and Fast Gradient Method adversarial regularization. The method changes only the training objective and procedure; inference uses the same single base-size encoder and linear classifier as the corresponding label-smoothed baseline. We evaluate the method on two CSL-based tasks: an abstract-to-discipline task with 67 labels and a title-to-category task with 13 categories. On the primary abstract task, AutoTail-BSFGM improves validation and lockbox accuracy under both Chinese RoBERTa-WWM and MacBERT-base. With MacBERT-base, validation accuracy increases by 0.83 percentage points and lockbox accuracy by 0.49 points, with a pooled paired McNemar signal on validation (p = 0.023). On the title task, the method improves validation accuracy by 0.70 points and validation balanced accuracy by 2.64 points; lockbox accuracy is approximately neutral while lockbox balanced accuracy improves by 1.22 points. The results support a bounded contribution: AutoTail-BSFGM improves class-balance-sensitive behavior and yields consistent gains for abstract-based scholarly classification, without uniformly improving every metric on every split.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑