【作者】 李银花；

【导师】 张继福；

【作者基本信息】 华北工学院 ， 计算机应用技术， 2004， 硕士

【摘要】 数据库技术已经从原始的数据处理，发展到开发具有查询和事务处理能力的数据库管理系统。进一步的发展导致越来越需要有效的数据分析和数据理解工具。粗糙集理论正是一种处理不确定、不一致数据的数学工具。本文的主要研究成果有以下三个方面：1.提出了一种基于区分矩阵的属性约简算法ARDM 通过对基于区分矩阵的属性约简算法进行分析，找出影响时间效率的因素并将命题演算中的吸收率用于构造区分矩阵的过程中，从而去掉了在区分函数中不起作用的“重复”元素，提高了属性约简的效率。2.提出了一种增量式属性约简算法针对实际问题中数据库中的数据是不断变化的这一情况，以粗糙逻辑为基础，针对新加入的对象相对于原来的极小决策算法而言是全新的这种情况，提出了一种增量式属性约简算法，从而避免每次从庞大的原始决策表开始约简，实现了对原极小决策算法的更新与维护，提高了属性约简的效率。3.依据上述属性约简算法，采用Visual FoxPro 6.0数据库管理系统，设计与实现了恒星光谱数据自动分类系统，实现了恒星光谱分类规则的自动提取。更多还原

【Abstract】 Database Technology has already been developed from original data processing to Database Management System, which can inquire and process transaction. With the further development, the efficient data analysis and data understandability tools are more and more needed. Rough sets is just a mathematical tool to deal with imprecise and inconsistent knowledge. This paper shows three aspects as follow:1. An algorithm of attribute reduction based on discernibility matrix ----ARDM is presented.Through analyzing to the algorithm of attribute reduction based on discernibility matrix, the facts that affect time efficiency are found, and the absorptivity in the proposition calculation is used to the process of constructing the discernibility matrix, then the effectless repeated elements are deleted, accordingly the efficiency of attribute reduction is improved.2. An dynamic algorithm of attribute reduction is presented. In fact, data are always changing in database, so a dynamic algorithm of attribute reduction based on rough logic is presented, which can get new minimum decision algorithm based on the original one when new object is added. It can avoid reduction from large original decision table, update and vindicate the original algorithm, and improve the efficiency of attribute reduction.3. According to the algorithm of attribute reduction, auto-classification system about stellar spectrum data is designed and implemented through using Visual FoxPro 6.0, which automatically extract rules from stellar spectrum.更多还原