大数据环境下置信规则库推理模型的优化与应用研究

How to use the massive data to make decision under big data environment is an important yet frontier research issue that needs to be investigated urgently using quantitative and qualitative approaches by modern management and decision support system theory. This project aims to make a deep and systematic study on the problems of structure optimization, decision response and information overload of belief rule-base (BRB) inference models from the perspective of big data, with the goal of optimizing the number of belief rules, the time complexity and the joint decision-making ability of belief rule-base inference models. The optimization theory and methods for belief rule-base inference models will be suggested. Applications to emergency response and medical service will be investigated using belief rule base inference models. The main research contents include establishing the decision-making-unit matrix based on belief rule-bases, determining the effectiveness of the decision making units based on parallel computing, proposing methods for rule reduction in belief rule-base under considering its completion, proposing methods for creating the index framework of belief rule-base, proposing rule retrieval methods of belief rule-base for optimizing decision response, proposing methods for building belief rule-bases with diversity, creating ensemble learning mechanism for belief rule-base inference models, and constructing application models for emergency response and clinical practice guidance. The research outcomes have important theoretical and practical significance to the enrichment and development of the theory of management and decision support system as well as providing decision supports and guidance for government departments and medical institutions.

大数据环境下如何有效利用大规模数据作决策是现代管理与决策支持系统理论迫切需要研究的前沿课题。本项目从大数据视角对置信规则库推理模型的结构优化问题、决策响应问题和信息过载问题进行深入系统地研究，以优化置信规则库中的规则数量、置信规则库推理的时间复杂度以及置信规则库推理模型的联合决策能力为目的，提出置信规则库推理模型的优化理论与方法，并将其应用于应急响应和医疗服务等领域。主要研究内容有：基于置信规则库决策单元矩阵的构造方法，基于并行计算的决策单元有效性判别方法，兼顾完备性的置信规则库规则约减方法，置信规则库索引框架的构建方法，具有快速决策响应的置信规则库规则检索方法，具备多样性的置信规则库构建方法，置信规则库推理模型的集成学习机制，突发事件应急响应问题和临床诊疗指导问题的应用模型的构建与优化。研究成果对发展和完善管理与决策支持系统理论、为政府部门与医疗机构提供决策支持具有重要的理论和实际意义。

大数据环境下如何有效利用大规模数据作决策是现代管理与决策支持系统理论迫切需要研究的前沿课题。本项目从大数据视角对置信规则库推理模型的结构优化问题、决策响应问题和信息过载问题进行深入系统研究，以优化置信规则库中的规则数量、置信规则库推理的时间复杂度以及置信规则库推理模型的联合决策能力为目的，提出置信规则库推理模型的优化理论与方法，并将其应用于应急响应和医疗服务等领域。主要研究内容有：基于置信规则库决策单元矩阵的构造方法，基于并行计算的决策单元有效性判别方法，兼顾完备性的置信规则库规则约减方法，置信规则库索引框架的构建方法，具有快速决策响应的置信规则库规则检索方法，具备多样性的置信规则库构建方法，置信规则库推理模型的集成学习机制，突发事件应急响应问题和临床诊疗指导问题的应用模型的构建与优化。本项目所取得的重要成果包括：共计完成学术论文124篇，其中SCI期刊论文72篇（含18篇为计算机科学学科Top期刊）。这些成果被SCI收录的高水平论文引用384次，在中国知网上被其他学者引用93次；获福建省社会科学优秀成果奖1次；培养学科带头人2人、硕博士研究生48人、博士后2人，其中已取得硕博士学位的研究生26人，获福州大学优秀博士学位论文6人、获批国家基金项目10项（9人）；与西班牙哈恩大学、英国阿尔斯特大学、新加坡国立大学、新西兰奥克兰大学紧密合作，出国合作交流8人，合作发表SCI期刊论文11篇；上述成果远超“完成学术论文16-20篇，其中SCI论文8-12篇；培养硕博士研究生8-10人”的原定目标。研究成果对发展和完善管理与决策支持系统理论、为政府部门与医疗机构提供决策支持具有重要的理论和实际意义。