社会网络环境下双驱动DEMATEL的群智知识融合应急决策方法

Group intelligence fusion emergency decision-making method with dual-driven DEMATEL in a social network environment

针对多属性应急群决策中决策属性缺少数据支持和公众难以参与决策过程的问题,提出了一种使用信任网络计算专家权重和融合公众知识与专家知识的双驱动模型的应急决策方法。首先,考虑传统模型只能依赖主观经验的不足,通过分析社交媒体中的文本数据来获得公众意见,并使用词频-逆文档频率算法(Term Frequency-Inverse Document Frequency, TF-IDF)提取意见中的关键信息,以公众大数据来获得数据驱动因素,同时,通过决策专家提供的专家知识,为决策过程提供知识驱动因素,构成双驱动的决策实验室分析法(Decision Making Trial and Evaluation Laboratory, DEMATEL)模型来建立评价属性体系,模型中影响因素的相互作用程度由公众大数据与专家评价共同决定,以得到公众知识数据与专家知识评价融合的结果;其次,使用社会网络表示专家之间的信任关系与信任强度,并通过Louvain算法对专家进行聚类,通过社会网络中节点的度中心性与接近中心性,得到个体的权重进而计算出各聚类权重,使用直觉模糊加权平均算子(Iterative Fuzzy Weighted Averaging, IFWA)结合决策偏好与属性权重,通过得分函数计算备选方案的得分,并依据得分结果对方案排序以得到最优的解决方案;最后,结合“7·20”郑州市突发暴雨案例证明了本方法的可行性和有效性。

To address the insufficient data support for decision attributes in multi-attribute emergency group decision-making and the challenges associated with public participation in the process,this paper introduces an innovative emergency decision-making method.The objective is to combine public knowledge and expert insights through the application of a trust network for determining expert weights and implementing a dual-driven model.The conventional models often depend solely on subjective experience,leading to limitations in their effectiveness.To surpass this constraint,the suggested method integrates a dual-driven Decision Making Trial and Evaluation Laboratory(DEMATEL)model.This model utilizes the analysis of text data from social media to capture public opinions effectively.Utilizing the Term Frequency-Inverse Document Frequency(TF-IDF)algorithm,essential insights are extracted from these opinions,facilitating the identification of data-driven factors through the extensive pool of public big data.Concurrently,decision-making experts expert knowledge plays a pivotal role in formulating knowledge-driven factors.By integrating these elements,the dual-driven DEMATEL model constructs an evaluation attribute system that merges public knowledge data and expert knowledge assessment.This methodology guarantees a more inclusive and equitable decision-making process.To depict the trust dynamics among experts,a social network is utilized.The Louvain algorithm is employed to cluster the experts,taking into account the connections and interactions within the network.The weighting of individuals is established by assessing their centrality degree and proximity to the central nodes in the social network.Moreover,the weights of each cluster are computed,factoring in the expertise and reliability of the experts.Additionally,decision preferences and attribute weights are merged through the Iterative Fuzzy Weighted Averaging(IFWA)operator,enabling a holistic evaluation of the alternatives.The suggested approach utilizes a scoring function to determine the scores of the alternatives effectively.Using the acquired scores,the programs are ranked to pinpoint the optimal solution.By taking into consideration expertise,trustworthiness,decision preferences,and attribute weights,the proposed method delivers a more objective and resilient decision-making process.To validate the practicality and efficacy of the method,it is tested in a case study concerning the“7.20”Zhengzhou City sudden rainstorm.The outcomes are then compared and scrutinized alongside other prevailing methods documented in the literature.