项目名称: 多标记数据的粒计算理论与算法研究
项目编号: No.61272095
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 李德玉
作者单位: 山西大学
项目金额: 81万元
中文摘要: 多标记数据是密切关联与Web信息处理、文本分类、生物信息学、视频语义标注、信息检索等众多应用领域的一类重要数据类型。粒计算是当前人工智能,特别是智能信息处理领域最为活跃的研究方向,对大规模复杂数据的建模、分析与计算任务具有重要意义。本课题旨在研究多标记数据粒计算理论与算法。主要内容包括:(1)数据的极小表示拓扑结构学习;(2)不确定性度量与属性约简算法;(3)数据的覆盖聚类粒化;(4)多标记数据分类优先性;(5)数据的概念结构及其可视化。系统地发展多标记数据建模、分析与计算的粒计算理论与方法,开发一个可用于多标记数据建模与分析的实验系统。本项目研究成果将丰富粒计算理论,拓展粒计算的数据适用范围,探索复杂信息处理的新途径,对数据挖掘和知识发现的研究有重要的理论意义和应用价值。
中文关键词: 粒计算;多标记数据;粗糙集;概念格;
英文摘要: Multi-lable data is an important data type which is associated with many application fields such as Web information processing, text categorization, bioinformatics, video semantic annotation, information retrieval. Granular computing is one of the most active research fields in artificial intelligence, especially in intelligent information processing. It has a special significance for modeling, analysis and computation based on large-scale complex data. This project devotes itself to the theory and algorithm research on multi-lable data. The main contents include: (1) minimal representation topology structure learning of data; (2) uncertainty measure and attribute reduction algorithm; (3) granulating data based on coverage clustering; (4) classification priority of multi-lable data; (5) data concept structure and its visualization. This project will systematacially develop the theory and method of granular computing for modeling, analysis and computation of multi-label data, and exploit an experimental system for multi-label data modeling and analyzing. The research results of this project will enrich the granular computing theory, expand the data scope for it, explore the new way for complex information processing, and has the very important theoretical significance and practical application value for data mini
英文关键词: Granular computing;Multi-lable data;Rough set;Concept lattice;