基于组合赋权的暴恐转向风险预测研究

研究群体性事件演变过程中的冲突升级引发暴恐转向的风险，对维护国家安全和社会稳定具有重要意义。本文提出基于词频-逆文档频率(Term Frequency-Inverse Document Frequency, TF-IDF)及三角模糊数(Triangular Fuzzy Number, TFN)的组合赋权法，并融合贝叶斯网络(Bayesian Network, BN)构建风险转向预测模型。首先，通过分析促使事件态势演变的影响因素，建立暴恐转向指标体系，利用TF-IDF方法对社交媒体文本数据中的代表性高频词汇进行筛选和分类，计算对应指标的客观权重。其次，结合TFN和德尔菲法设计调查问卷，采用全积分值法对专家小组评判结果进行解模糊化处理，得到指标的主观权重。再次，根据博弈论组合赋权法得到综合权重，融合BN概率推理模型，预测发生暴恐转向的概率。最后，利用反向推理得出影响因素的后验概率，分析引发暴恐转向的主要因素，为相关部门采取针对性措施提供理论参考。实例结果显示暴恐转向风险的预测概率为81%,与实际事件态势发展情况相符，验证了本文模型对群体性事件态势发生暴恐转向风险具有可行、准确的预测。

Research on Risk Prediction of Violent Terrorist Turn Based on Combination Weighting

Studying the risk of violent terrorist turn during the evolution of mass incidents is of great significance for maintaining national security and social stability.A risk turn prediction model that integrates the combined weighting method based on Term Frequency-Inverse Document Frequency (TF-IDF) and Triangular Fuzzy Number (TFN),and Bayesian Network (BN) were proposed.Firstly,by analyzing the influencing factors that promote the evolution of the event situation,the index system of violent terrorist diversion were established.The TF-IDF method was used to screen and classify the representative high-frequency words in the social media text data,and the objective weight of the corresponding index was calculated.Secondly,the questionnaire was designed by combining triangular fuzzy numbers and Delphi method.According to the evaluation results of the expert group,the total integral value method was used to defuzzify the investigation results and the subjective weights of the indicators were obtained.Thirdly,according to the combination weighting method of game theory,the comprehensive weight was obtained,and the probability reasoning model of Bayesian network was integrated to predict the probability of violent terrorist turn.Finally,the posterior probability of influencing factors was obtained by reverse reasoning,and the main factors that lead to the turn of violent terrorist were analyzed,which provided theoretical reference for relevant departments to take targeted measures.The example results show that the prediction probability of the risk of violent terrorist turn is 81%,which is consistent with the development of the actual event situation,which verifies that the model in this article has a feasible and accurate prediction of the risk of violent terrorist turn in the situation of group events.