一种基于约束概念格的恒星光谱数据自动分类方法  

Automatic Classification Method of Star Spectrum Data Based on Constrained Concept Lattice

作　　者：张继福[1] 马洋[1] 

机构地区：[1]太原科技大学计算机科学与技术学院,山西太原030024

出　　处：《光谱学与光谱分析》2010年第2期559-562,共4页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(6057307560773014);山西省回国留学人员科研项目(2009-77)资助

摘　　要：概念格是数据分析与知识提取的一种有效形式化工具,约束概念格是一种新的概念格结构,具有构造的时空复杂性低,所提取知识的实用性和针对性强等特点。针对海量恒星光谱自动分类任务,依据约束概念格结点外延与训练样本数据集等价划分之间的关系,通过引入外延支持度和划分支持度的概念,提出了一种基于约束概念格的恒星光谱数据分类规则挖掘方法,并采用SDSS恒星光谱数据集,实验验证了由该方法所提取出的分类规则具有较高的分类效率和分类正确率,从而为海量的恒星光谱数据自动分类提供了一种有效方法。Concept lattice is an effective formal tool for data analysis and knowledge extraction. Constrained concept lattice, with the characteristics of higher constructing efficiency, practicability and pertinency, is a new concept lattice structure. For the au- tomatic classification task of star spectrum, a classification rule mining method based on constrained concept lattice is presented by using the concepts of partition and extant supports. In the end, the experimental results validate the higher classification effi- ciency and correctness of the method by taking the star spectrum data as the formal context, so that an effective way is provided for the automatic classification of massive star spectrum.

关 键 词：恒星光谱 约束概念格 分类规则 划分支持度 外延支持度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]