联系我们

论文查询结果	返回搜索

论文编号：	12369
作者编号：	1120180971
上传时间：	2021/6/9 18:37:45
中文题目：	面向去碎片化的公共政策知识聚合研究
英文题目：	Research on Knowledge Aggregation of Public Policy for Defragmentation
指导老师：	王芳
中文关键字：	公共政策；去碎片化；知识聚合；政策网络；知识图谱
英文关键字：	Public policy; Defragmentation; Knowledge aggregation; Policy network; knowledge graph
中文摘要：	公共政策碎片化是我国持续推进国家治理体系和治理能力现代化不得不面对的深层次难题，既是政府治理碎片化的结果，反过来又强化了政府治理的割裂与碎片化程度。中国是一个“政策大国”，加之由中国的国情所决定，中国政策网络的复杂程度和碎片化程度可能远超过其他国家和地区。伴随着公共服务的深化和政策文本的流变，政策文件快速增长，政策碎片化问题愈加突出，亟待通过对大规模公共政策的深入研究，运用大数据、深度学习、知识图谱等技术，智能识别、预警与消解政策系统中的碎片化问题，为大数据时代提升政府的整体性治理能力和现代化治理水平提供理论与技术支撑。 目前现有研究缺乏对大规模公共政策文本语料的深层关联与有效聚合，这一技术瓶颈导致政策文本资源知识复用困难，难以实现公共政策去碎片化目标。基于此，本文以公共政策碎片化为问题导向，综合运用情报学、计算机科学、公共管理等多学科理论与方法，探索了如何通过公共政策关联聚合这一技术路径实现公共政策去碎片化目标。首先对公共政策碎片化进行了问题剖析与理论建模，然后提出公共政策知识聚合技术框架，基于大规模政策文本语料构建了超大政策网络，实现了公共政策的时空关联、知识聚合与关系推理，最后设计了公共政策去碎片化技术方案。具体而言，本文的主要研究工作包括： 第一，通过内容分析法和多案例文本分析，对公共政策碎片化进行了问题剖析和理论建模。研究基于文本视角，将公共政策碎片化科学划分为八种主要类型：政出多门、政策打架、政策冲突、政策变通、政策脱节、政策孤岛、政策分散、政策疏漏，将碎片化语义特征分为六种主要类型：政策主体碎片化、政策客体碎片化、政策目标碎片化、政策依据碎片化、政策工具碎片化、政策内容碎片化。 第二，重点探讨了公共政策多粒度知识聚合的实现方法，提出基于要素转化思想的公共政策知识聚合流程。以“COVID-19”政策为样本数据，通过知识表示方法体系构建、多粒度知识识别、知识关联与对齐等关键步骤，实现了政策文本的知识聚合。在此基础上，从聚合对象、聚合规则、聚合层次三个方面对公共政策知识聚合模式进行了详细讨论。实验表明，研究构建的公共政策知识图谱能够清晰展示政策主体、政策客体、政策目标、政策依据、政策主题等不同维度下的政策族谱，在多维分析、计算效率、可扩展性、可视化等方面表现优异。 第三，基于15.7万余篇公共政策文本语料，经过数据预处理、知识表示、知识抽取、知识对齐、知识存储、知识聚合等步骤，构建了具有91.5万节点、373.2万关系的超大政策网络。探索实现了公共政策时空关联的界面化分析，基于图算法设计了公共政策碎片化识别规则，并选取两个重大突发公共危机的典型事件“SARS”和“COVID-19”，对构建的网络数据和识别算法进行了规模化应用验证。 第四，提出基于知识聚合的公共政策去碎片化智能应用方案，主要包括两个方面。一是面向政策查询需求场景下的知识应用，包括公共政策碎片化信息整合、政策网络分类展示等；二是面向政府现实场景的精准知识服务，包括政策学习与政策移植、政策冲突预警以及政策知识智能处理等。 本文的主要改进和创新体现在以下四个方面：第一，基于政策文本视角构建了公共政策碎片化分类体系与语义特征分析框架，为研究公共政策去碎片化智能处理技术提供理论依据。第二，设计了公共政策知识聚合方法与流程，实现了对公共政策的语义识别和关联聚合，极大便利了异常点和异常关系的检测，为碎片化政策的协调、关联、整合和冲突识别提供了有效的技术支撑。第三，对大规模公共政策文本资源开展深层次语义关联，构建了超大政策网络，实现了公共政策的时空关联与知识聚合，多维度揭示复杂政策系统，展现了政策间的高度关联性，为推动政策科学更加有效地解决社会复杂问题提供基础。第四，提出了基于知识聚合的公共政策去碎片化技术解决方案，为有效破解政府碎片化治理难题提供技术路径与方法参考。 图47幅，表50个，参考文献290篇。
英文摘要：	Fragmentation of public policy is a deep-seated problem that China has to face when it continues to promote the modernization of its national governance system and governance capabilities. It is not only the result of fragmentation of government governance, but also strengthens the degree of fragmentation of government governance. China is a "big policy country", the complexity and fragmentation of China's policy network may far exceed that of other countries and regions determined by China's national conditions. With the deepening of public services and the evolution of policies, policy documents are growing rapidly, and the problem of policy fragmentation is becoming increasingly prominent. It has become a deep-seated problem that China has to face in the process of deepening reform. It is urgent to conduct in-depth research on large-scale policy texts, and use big data, deep learning, knowledge graph and other technologies to intelligently identify, early warn and eliminate the problems of fragmentation in the policy system, in order to provide theoretical and technical support for improving the government's overall governance ability and modern governance level in the era of big data. At present, the existing research lacks of deep association and effective aggregation of large-scale public policy text corpus, which makes it difficult to reuse the knowledge of policy text resources and achieve the goal of public policy de-fragmentation. Based on this, this paper aims to realize the defragmentation of public policy through the technical path of semantic association and knowledge aggregation of public policy, guided by the fragmentation of public policy, comprehensively uses the theories and methods of information science, computer science, public management and other disciplines, analyzes the problem and theoretical modeling of the fragmentation of public policy, and puts forward the technology of knowledge aggregation of public policy, realizes the spatiotemporal association, knowledge aggregation and relational reasoning of public policy, constructs a large-scale policy network based on large-scale policy text corpus, and designs a technical solution for public policy defragmentation. Specifically, the main research work of this paper includes: First, through the content analysis and multi case text analysis, the fragmentation of public policy is analyzed and modeled theoretically. Based on the text perspective, this paper divides the fragmentation of public policy into eight main types: divided policies from various sources, policy fights, policy conflict, policy flexibility, policy disconnection, policy Island, policy decentralization, policy omissions. The semantic features of fragmentation can be divided into six main types: policy subject fragmentation, policy object fragmentation, policy objective fragmentation, policy basis fragmentation, policy tool fragmentation and policy content fragmentation. Secondly, this paper focuses on the realization of multi-granularity knowledge aggregation of public policy, and proposes the process of knowledge aggregation of public policy based on the idea of factor transformation. Taking "COVID-19" policy data as the sample, the knowledge aggregation of policy texts is realized through the key steps of knowledge representation system construction, multi-granularity knowledge recognition, knowledge association and alignment. On this basis, this paper discusses the aggregation model of public policy knowledge in detail from three aspects: aggregation object, aggregation rule and aggregation level. Experiments show that the public policy knowledge map can clearly show the policy genealogy of different dimensions, such as policy subject, policy target group, policy objective, policy basis, policy theme, and it is excellent in multidimensional analysis, computational efficiency, scalability, visualization, etc. Third, based on more than 157,000 public policy text corpora, through data preprocessing, knowledge representation, knowledge extraction, knowledge alignment, knowledge storage, knowledge aggregation and other steps, a super-large policy network with 0.915 million nodes and 3.732 million relationships has been constructed. This paper explores and realizes the interface analysis of the temporal and spatial correlation of public policies, designs the identification rules of public policy fragmentation based on the graph algorithm, and selects two typical events of major public emergencies “SARS” and “COVID-19” as examples, Network data and recognition algorithms have been verified for large-scale applications. Fourth, a public policy defragmentation smart application scheme based on knowledge aggregation is proposed, which mainly includes two aspects. The first is knowledge application in the context of policy query needs, including the integration of public policy fragmentation information, and the faceted and classified display of policy networks; the second is accurate knowledge services for government real-world scenarios, including policy learning and policy transplantation, policy conflict warning, and Intelligent processing of policy knowledge, etc. The innovations of this article mainly include: First, based on the perspective of policy text, a public policy fragmentation classification system and semantic feature analysis framework are constructed to provide a theoretical basis for the study of public policy defragmentation intelligent processing technology. Second, the method and process of public policy knowledge aggregation are designed to realize the semantic recognition and association aggregation of public policies, which greatly facilitates the detection of abnormal points and abnormal relationships, provides effective technical support for the coordination, association, integration and conflict identification of fragmented policies. Third, through the deep semantic association of large-scale public policy text resources, a super-large policy network is constructed, which realizes the temporal and spatial correlation and knowledge aggregation of public policies, reveals complex policy systems in multiple dimensions, and demonstrates the high degree of relevance between policies, provides a basis for promoting policy science to solve complex social problems more effectively. Fourth, a technical solution for the defragmentation of public policies based on knowledge aggregation is proposed to provide a technical path and method reference for effectively solving the problem of government fragmentation governance. 47 pictures, 50 tables, 290 references.
查看全文：	预览  下载(下载需要进行登录)