2026-10-13T23:27:40Z https://www.scidb.cn/oai

10.57760/sciencedb.44911 2026-08-03T11:12:24Z

2026-08-03 Supporting Data for Research on Zero-Shot Cybersecurity Entity Extraction Method Based on Multi-Agent Framework doi:10.57760/sciencedb.44911 en This dataset provides the supporting experimental materials for the paper “Research on Zero-Shot Cybersecurity Entity Extraction Method Based on Multi-Agent Framework.” It contains cleaned source code for the CADRE framework and nine baseline methods, prompt templates, model and environment configurations, and eight groups of aggregate results covering comparative experiments, ablation studies, parameter sensitivity, efficiency analysis, and cross-model robustness evaluation. Results are provided in CSV, XLSX, and Markdown formats, together with file manifests, checksums, and reproducibility instructions. The experiments are based on the third-party public APTNER dataset. The original APTNER files are not redistributed; instead, the package provides the pinned official repository version, paper DOI, SHA-256 checksum of the test file, and download instructions. The evaluation uses 1,256 entity-containing sentences from the APTNER test set, yielding 3,383 entity chunks. API credentials, per-sample prediction files, and complete runtime logs are not included. Cybersecurity; Named Entity Recognition; Multi-Agent System; Zero-Shot Learning; Cyber Threat Intelligence; CADRE Liu Jingyuan PUBLIC https://creativecommons.org/licenses/by-nc-sa/4.0/ dataset Science Data Bank