Abstract: Transformation from military document into corresponding map can usually involve non-trivial manual work while automatic document-map transformation system may significantly contribute to the automation of military commanding. Based on the HNC theory and its semantic model, the paper proposes and implements a system that can automatically accomplish such a transformation. By automatically retrieving the name of a troop, temporal and spatial information, together with the result of analysis of sentence category, the proposed approach can fill the contextual framework of a sentence group, whose information can then be easily transferred into map information. Extensive experiments of real military document illustrate that the proposed method can reach a higher rate of precision and recall and hence provide a reliable approach for automatic document-map transformation.

Key words: semantic analysis, analysis of sentence category, military document, map depicting, document-map transferring

摘要： 军事文书转换为要图通常需要耗时费力的人工工作，文图的自动转换将大大促进军事指挥的自动化水平。基于HNC理论及其语义模型提出并实现了一个文图自动转换系统，通过对军事文书中部队番号的自动识别、时间地点要素的自动获取，以及语句的句类分析可以获得句群文本的语义情景框架，根据情景框架中的要素可以直接转换为要图标绘所需的信息。实验表明军事文书的文本自动分析系统具有较高的准确率和召回率，为实现军事文书的文图自动转换提供了一种可靠方法。

关键词: 语义分析, 句类分析, 军事文书, 要图标绘, 文图转换