大数据背景下基于知识图谱的电子产品回收政策环境效应与经济性分析方法

Effective end-of-life management of electronic products is becoming a pressing concern, as well as a difficult challenge. Due to lack of panoramic and integrated data and knowledge, it is impossible to conduct a panoramic and stereoscopic analysis on both environmental and economic perspectives of the policies for regulating and recycling end-of-life electronic products. Consequently, the effectiveness of these management policies is compromised, resulting in low recycling rates, resources in idle and serious pollutions. Since electronic products are manufactured in high quantities, of various types and evolve enormous users, the related life cycle data has the characteristics of big data. Based on knowledge graph (a type of knowledge network), this project investigates the theory and methodology for analyzing the environmental and economic aspects of the end-of-life management policies of electronic products, panoramically and stereoscopically, with the objective to promote the effectiveness of such policies. First, based on data of product families and recycling cycles, a panoramic knowledge graph of end-of-life management of electronic products is constructed. Second, to address the challenges of modeling under the context of big data, adaptive knowledge extraction and matching methods are proposed for automatically acquiring and integrating relevant knowledge. Third, based on the panoramic knowledge graph of end-of-life management of electronic products, knowledge-based reasoning mechanism is investigated for supporting the analysis of the effectiveness of the management policies from both environmental and economic perspectives, and for providing theoretical fundamentals and practical tools to overcome the extant barriers in recycling electronic products. This project enriches the theory and application scenarios of knowledge management, extending the theoretical frontier and applications of knowledge management (knowledge graph) to the analysis in public management and environmental management; it is an interdisciplinary study of great theoretical and practical importance.

电子产品回收的政策研究困难、急迫。由于缺少全景信息集成分析，难以对回收政策进行全景立体的环境效用与经济性分析，从而导致政策失效，回收率较低且资源闲置与环境污染严重。电子产品种类多、设计和回收技术多样化，环境影响因素多，具有大数据的特征。基于知识图谱，本项目研究大数据环境下的电子产品回收政策全景立体的环境效用与经济性分析理论与方法，为政策的细化和实施提供决策依据。首先，建立基于电子产品族和回收全周期的全景知识图谱。然后，研究大数据背景下知识自适应抽取与关联方法，以实现知识集成。基于全景知识图谱开展面向电子产品回收政策的知识推理机制研究，对相关政策的技术实施途径的环境效用与经济性进行推理性分析，为解决电子产品回收政策实施难的问题提供理论支撑、方法和工具。本项目丰富了知识管理理论和应用场景，同时又是知识管理理论在公共管理（环境管理）政策分析领域的理论拓展，具有理论和实践双重意义的学科交叉研究。

缺少有效的信息与知识以支撑针对电子产品回收管理的全景式环境效用与经济性分析，从而导致电子产品回收政策失效，回收率较低且资源闲置与环境污染严重。基于获取整理的电子产品回收大数据，本项目研究首先对知识图谱的建模理论与实现过程进行了系统调研与实验尝试，分析指出了知识图谱特征建模中冷启动问题与长尾问题的原因，即由数据多样性与分布不均而引起的少样本问题。通过挖掘元知识以进行元匹配、元关联与元学习，提出了两种MMM算法以解决少样本分类问题。基于公开数据集与电子回收大数据，两种算法的有效性与优越性得以验证。进一步，结合LSTM等深度学习神经网络设计多种关注机制进行小样本学习与跨领域关联，实现了知识自适应抽取与关联，提出了基于大数据的知识图谱自学习建模体系并构建了电子产品回收全景知识图谱。基于电子产品回收全景知识图谱，构建了基于知识图谱的政策环境效应与经济性的系统分析方法论，对电子产品回收管理的相关典型政策、策略与场景进行了多角度探索。应用基于知识图谱的知识推理机制，对电子产品回收管理的新型场景（如城市垃圾处理综合体）与政策导向（如鼓励锂电池梯次利用体系研发），对废弃电子产品（如家电与锂电池）的处理路径进行了全景式环境效应与经济性的综合评估，量化了系列影响因素（如电器产品迭代、电力价格放开等）对目标系统表现的影响，考虑了处理路径的社会可接受性。结合知识图谱的嵌入关联与所提出的图推理算法，对政策响应实验流程进行了改造，并通过政策响应实验探索了环境政策宣传—环保意识—支付意愿之间的复杂关联，揭示了多种心理作用（如喜悦与预防定向）的影响。基于上述理论与方法，本项目对新兴电子产品（如AR眼镜）的技术发展、产品功能与结构，设计了兼具环境与经济可行性的回收路径。本项目研究丰富了知识管理理论与方法，拓展了其在公共管理与环境管理的交叉发展，解决电子产品回收难提供了理论、方法和工具，具有理论和实践双重意义。

内容获取失败，请点击重试