自然语言处理数据集|军事新闻分析数据集

CMNEE是一个面向中文军事新闻事件抽取的大规模标注数据集，包含17,000份文档和29,223个事件，涵盖8种事件类型和11种论元角色。数据集采用两阶段多轮次标注策略，结合触发词字典预标注与领域专家人工审核，确保高质量与专业性。其规模大、标注精细，适用于军事领域事件检测、信息抽取与自然语言理解等任务，可广泛应用于情报分析、自动摘要与知识图谱构建等场景。

CMNEE is a large-scale annotated dataset dedicated to Chinese military news event extraction. It contains 17,000 documents and 29,223 events, covering 8 event types and 11 argument roles. The dataset adopts a two-stage and multi-round annotation strategy, combining pre-annotation via trigger word dictionaries and manual verification by domain experts to ensure high quality and professional rigor. With large scale and fine-grained annotations, it is suitable for tasks such as military domain event detection, information extraction and natural language understanding, and can be widely applied in scenarios including intelligence analysis, automatic summarization and knowledge graph construction.

搜集汇总

背景与挑战

背景概述

CMNEE是一个大规模、文档级的中文军事新闻事件抽取数据集，包含17,000份文档和29,223个事件，覆盖8种事件类型和11种论元角色。该数据集采用触发词字典预标注与领域专家人工审核相结合的两阶段多轮次标注策略，确保数据的高质量与专业性，适用于事件抽取模型的研究和评估。