Abstract:

This work discusses the connotations of related concepts and the evolution process in the field of strategic intelligence warning and prediction. In consideration of national and military conditions， a brand new definition of strategic intelligence warning and prediction is proposed. Specific to the requirements and characteristics， a basic process and intelligent methodology of strategic intelligence warning and prediction activity are proposed. This work summarizes the core ideas of different intelligent methods， introduces the main available technologies and models， which are useful in guiding and providing reference to strategic intelligence researchers in terms of tool selection and approaches formulation. Cognitive driven methods have distinct advantages in interpretability and interactivity， while data driven methods are suitable for large-scale processing. Toachieve optimal performance in strategic intelligence warning and prediction targets， the two methodologies should be used in combination. These intelligent methods need to be improved to solve the key problems of man-machine cognitive difference， multi-granularity knowledge fusing， hidden clue recognition and evidence confirming， to improve the comprehensive， accuracy and reliability.

Key words: strategic intelligence, warning and prediction conception, warning and prediction process, intelligent methodology, solution framework

摘要：

系统性探讨了国内外战略情报预警预测相关概念内涵及其演变过程，给出了契合新时期我国情与军情的战略情报预警预测定义。针对预警预测要求与活动特点，提出了战略情报预警预测的基本环节与流程框架，归纳总结了实施战略情报预警预测的认知驱动类和数据驱动类智能化方法的基本思想和主要算法模型，能为战略情报分析领域研究人员开展研究提供工具选择、方法途径等方面的指导与参考。认知驱动类方法有可解释性和交互性方面的优势，数据驱动类方法在大规模数据处理效率方面更佳，两者结合在人机协同式的复杂战略情报预警预测分析任务中能发挥更好的效能。未来上述智能化方法将重点解决人机认知差异性度量、异构知识关联融合、弱隐微线索挖掘、证据链印证闭环等方面的问题，以满足预警预测领域的全面性、准确性、可信度等要求。

关键词: 战略情报, 预警预测概念, 预警预测流程, 智能化方法, 求解框架