摘要
在应用粗糙集理论对决策表进行数据处理时对数据约简算法进行了改进,以达到提高约简效率的目的.在进行属性约简时采用分辨矩阵的做法,但其缺点是生成中间过程的数据时需要耗费大量的资源,可以考虑在生成分辨矩阵时的循环过程中利用吸收律及时进行化简,在记录比较多的情况下对于降低空间利用率,提高运算速度是有效的.在对属性值进行约简生成规则时,针对生成等价类时间复杂度高的问题,不再对每一个节点约简时调用等价类生成函数,而是采用生成一次等价类,存储多次调用的办法,大大降低了时间复杂度,提高了运行效率.
The improvement to data reducing algorithm is made to improve the efficiency of data reducing while dealing with the data of decision table using Rough Set theory. When reducing data, this algorithm adopts the distinguishing matrix. But in this way the system requires a large quantity of extra disk space to save temporary data. The absorption law is employed to simplify them in time at cycle progress. It is very efficient in depressing the use of disk space and improving the operation velocity. In order to solve the problem of high time complexity in creating equal class, this algorithm adopts the way of creating equal class only once and saves it for transferring repeatedly instead of transferring the equal class function to every node, thus the time complexity can be lowered largely and the operation efficiency can be increased.
出处
《大连海事大学学报》
CAS
CSCD
北大核心
2004年第1期108-110,共3页
Journal of Dalian Maritime University
基金
辽宁省自然科学基金资助项目(2001101050).
