arXiv cs.CL· Maciej Jalocha, Johan Hausted Schmidt, William Michelsen·· 12 小时前AI 评分18
网络安全 NER 跨数据集泛化中的标签统一研究
Label Unification for Cross-Dataset Generalization in Cybersecurity NER
AI 导读
针对网络安全 NER 缺乏标准化标签、数据集难以合并的问题，研究对四个网络安全数据集做粗粒度标签统一，并用 RoBERTa 进行两两跨数据集评测。在所选的映射与训练设置下，统一训练在全部四个开发集上均不如对应的单数据集基线；共享权重的多头模型与基于图的迁移模型也未带来一致收益。
正文 · 原文
Abstract:The field of cybersecurity NER lacks standard-ized labels, making it challenging to combine datasets. We investigate label unification across four cybersecurity datasets to increase data resource usability. We perform a coarse-grained label unification and conduct pairwise cross-dataset evaluations using RoBERTa models. Qualitative analysis of predictions reveals errors, limitations, and dataset differences. To address unification limitations, we propose alternative architectures including a multi-head model and a graph-based transfer model. Under our selected mapping and training setup, unified training underperformed the corresponding single-dataset baselines on all four development sets. The multi-head model with weight sharing provides no consistent benefits over single-dataset baselines, and the graph-based transfer model shows no consistent benefits.
| Comments: | 7 pages, 15 tables, 1 equation |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2507.13870 [cs.CL] |
| (or arXiv:2507.13870v3 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2507.13870 arXiv-issued DOI via DataCite |
Submission history
From: Maciej Jalocha [view email]
[v1]
Fri, 18 Jul 2025 12:47:20 UTC (18 KB)
[v2]
Sun, 31 Aug 2025 11:08:10 UTC (18 KB)
[v3]
Tue, 20 Oct 2026 13:09:10 UTC (30 KB)
来源：arXiv cs.CL · arxiv.org