机器学习与冲突预测——国际关系研究的一个跨学科视角

发布时间：2018-04-14 09:24

  本文选题：冲突预测 + 机器学习　； 参考：《世界经济与政治》2017年07期

【摘要】：通过机器学习来预测冲突正在成为当前国际关系研究领域的一个热议话题。但是从跨学科交叉研究的视角来看,计算机介入政治分析和国际关系研究并不是一个新现象,其间经历了从计算机模拟冲突场景到机器学习自动识别冲突模式的复杂变革历程。二者的共同点是都重视仿真社会互动情景与政治复杂演进过程,反对有关政治冲突现象的简单线性解释;但二者在研究取向上还是有着本质的不同,计算机模拟提倡基于特定社会理论的情景建模与逻辑推演,而机器学习则强调无特定社会理论支撑的特征识别与关联预测。有鉴于此,本文首先分析了计算机模拟与机器学习在冲突预测中的研究路径差异,然后重点阐述了无理论支撑下将机器学习应用于冲突预测之可能,并以2010—2016年印度恐怖袭击预测为例,实证检验了基于BP神经网络的机器学习在真实社会情景中的实际冲突预测效力,结果发现基于机器学习的冲突预测范式即使在没有特定社会理论支撑下,也具备一定冲突预测能力,并可产生新的冲突知识发现。但即便如此,作为一种跨学科交叉研究范式,机器学习介入冲突预测仍然面临重重困难。
[Abstract]:Prediction of conflict through machine learning is becoming a hot topic in the field of international relations.However, from the perspective of interdisciplinary research, computer involvement in political analysis and international relations research is not a new phenomenon, which has undergone a complex transformation from computer simulation conflict scene to machine learning automatic identification of conflict pattern.Their common point is that they attach importance to the simulation of social interaction scenarios and the complicated evolution process of politics, and oppose the simple linear explanation of the phenomenon of political conflict, but they still have essential differences in research orientation.Computer simulation advocates scenario modeling and logical deduction based on specific social theory, while machine learning emphasizes feature recognition and correlation prediction without the support of specific social theory.In view of this, this paper first analyzes the research path difference between computer simulation and machine learning in conflict prediction, and then focuses on the possibility of applying machine learning to conflict prediction without theoretical support.Taking the prediction of terrorist attacks in India 2010-2016 as an example, the paper empirically tests the effectiveness of machine learning based on BP neural network in actual conflict prediction in real social situations.The results show that the conflict prediction paradigm based on machine learning has a certain ability of conflict prediction even without the support of specific social theory and can produce new conflict knowledge discovery.But even so, as a cross-disciplinary research paradigm, machine learning involved in conflict prediction still faces a lot of difficulties.
【作者单位】： 对外经济贸易大学国际关系学院;
【基金】：国家社科基金青年项目“基于大数据驱动的外交决策模式创新与我国实践路径研究”(项目编号:15CGJ034) 对外经济贸易大学中央高校基本科研业务费专项资金资助(批准号:CXTD8-05)之阶段性成果
【分类号】：D80
，

本文编号：1748718