项目名称: 领域知识驱动的空间聚类及其人工免疫优化算法研究
项目编号: No.41201387
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 地理学
项目作者: 牛继强
作者单位: 信阳师范学院
项目金额: 24万元
中文摘要: 空间聚类是空间数据挖掘的重要组成部分,在地理信息科学领域具有重要的理论与应用研究价值。带有非空间属性的空间聚类分析是目前空间聚类研究的热点和难点问题。现有的空间聚类主要是对非空间属性进行扩展或对空间变量进行扩展,没有从本质上改变其聚类目标和聚类准则,同时在算法设计上缺少对子类属性内聚性的约束。本项目拟在对领域知识的分类和结构化解析研究的基础上,构建领域知识驱动的空间聚类形式化描述;根据空间聚类的特点和聚类任务的要求,重点研究领域知识驱动下的空间聚类模型,并对克隆选择优化算法的改进,提出可以应用于空间聚类的混沌免疫克隆选择聚类算法;设计并开发原型系统,以土地用途分区为例开展应用研究,验证模型的合理性和实用性。该研究有助于解决面向应用的空间聚类这一前沿问题,促进地理信息技术由数据驱动向模型驱动、知识驱动和决策支持转变。
中文关键词: 空间聚类;人工智能;土地利用规划;;
英文摘要: As an important issues in the domain of spatial data mining, spatial clustering has important theoretical and applied research value in the field of geographical information science. Because spatial clustering analysis has the non-spatial attributes, it can be considered to be a hot and difficult problem in current research. The existing research mainly focuses on the extended non-spatial attributes or spatial variables. However, this does not change the clustering objectives, as well as for the clustering criterion, in the spatial clustering. And in the aspects of algorithm design, the existing research lacks the study on the constraints of subclass attributes. Based on the classification of domain knowledge and the structured analytical study, the formal description of domain knowledge-driven spatial clustering is built. According to the characteristics of spatial clustering and the requirements of clustering task, the model of spatial clustering is focused on the domain knowledge-driven. And by improving clonal selection optimization algorithm, chaotic immune clonal selection clustering algorithm that can be applied to spatial clustering is proposed. In terms of the a prototype system designed in this project, the rationality and practical applicability of the algorithm of chaotic immune clonal selection clus
英文关键词: spatial clustering;artificial intelligence;land use planning;;