多尺度聚类知识挖掘与应用

多尺度聚类知识挖掘是数据挖掘的主要研究内容之一，也是研究热点，旨在发现潜在的多尺度集聚性分布模式。现有多尺度聚类知识挖掘大多局限于从空间尺度开展面向空间数据的聚类知识挖掘，缺乏从多尺度协同开展高性能的多类型聚类知识挖掘的理论与方法体系，难以满足用户需求。因此，本研究拟结合用户需求、数据特征和知识对尺度的强依赖性特征，开展多尺度（空间尺度、语义尺度、时间尺度）聚类知识挖掘研究，主要内容包括：（1）针对数据源尺度和聚类知识挖掘尺度与多尺度聚类知识的关联关系等问题，开展多尺度聚类知识挖掘的尺度认知；（2）研究多尺度聚类知识挖掘模式；（3）顾及空间属性、非空间属性以及时间属性，开展多尺度聚类知识挖掘与评价分析模型研究；（4）多尺度聚类知识挖掘系统研发，并以典型区域为例开展实例研究。本研究旨在探索复杂数据环境下满足用户需求的多尺度聚类知识挖掘理论与技术方法，为多尺度聚类知识挖掘提供新视角与新途径。

Multi-scale clustering, as one of main research points of data mining and knowledge discovery, is a research hotspot recently, which aims to mine the intrinsic clustering distribution patterns. Most of existed multi-scale clustering methods merely mine the clustering patterns of spatial attributes in the spatial scale, and these methods are difficult to meet the demands of users due to the lacking of comprehensively conducting clustering analysis in the spatial scale, sematic scale and time scale. In addition, the non-spatial attribute and time attribute have been ignored by existed multi-scale clustering methods. Hence, multi-scale clustering issues are analyzed by combining the user demand, data characteristics and clustering knowledge related to the scale response, and the scales include spatial scale, sematic scale and time scale. The analyzed clustering issues are as follows: (1) To the relationships between clustering knowledge and both data source scale and mining scale, the scale cognition of the multi-scale clustering is carried out; (2) Exploring the mining schema of the multi-scale clustering; (3) Exploring the clustering theory and methods by considering not only the spatial attributes, non-spatial attributes and time attributes, but also the multiple scales. (4) Developing the multi-scale clustering system and Carrying out practical applications. During the exploration of the above-mentioned issues, multi-scale clustering mining from new perspective are achieved and new approaches are proposed.