消费行业大数据驱动的融合微观-中观-宏观多层行为与目标的可预见全景式管理决策分析模型和方法

首先，对消费行业多源大数据中用户、产品、企业等微观个体行为，企业间竞争与协作关系等中观行业行为，及事件间因果关系等宏观行为，建立融合微观-中观-宏观多层行为及其内容与情感的知识图谱及其关联映射方法，以对消费行业大数据中碎片、稀疏和多样的行为及其内容和情感分布进行整体感知、关联与认知，以作为消费行业多源大数据可预见全景式管理决策知识表达与组织的载体；然后，以消费行业行为及其内容和情感的知识图谱为载体，以交互协同计算为手段，以消费行业多层行为之间的相互影响与相互作用关系为动力，研究融合多层行为与目标的消费行业运动轨迹的生成模型，为消费行业发展趋势的可预见全景式管理提供丰富的决策知识；继而，研究决策目标细化与演化方法，多样化全景式决策环境获取方法，管理决策知识优化机制及决策效果评估等；最后，研发食品行业多源大数据驱动的食品行业运行风险可预见全景式管理决策实例化应用平台，以验证研究的效果与可行性。

At first, this research innovatively constructs a three-layer knowledge base of behavior, content and emotion, i.e., microscopic entity knowledge base, mesoscopic industry knowledge base, and macroscopic event causality knowledge base, through mining multi-source big data of consuming industry, which can be considered as the cognitive representation and organization of the behavior, content and emotion hidden in the multi-source big data. Furthermore, combing spreading activation method and interaction among knowledge bases, a consuming industry learning model, which learns development trajectory and tendency of consuming industry, is built for the simulation and prediction of the consuming industry based on the constructed three-layer knowledge base. And then, with the three-layer knowledge base and its learning model in hand, this research is expected to study decision-making objective and its evolution, gain diversified and panoramic decision-making knowledge, and appraise decision-making effect. Finally, multi-source big data of food industry driven, a demonstration platform are designed and developed to realize the panoramic and customized management and decision-making.