Title :
Research on incremental calculation of maximum distribution reductions for inconsistent information systems
Author :
Chen, Hong-Tao ; Pan, Hong-Jun ; Liu, Jun
Author_Institution :
Sch. of Math., Phys., & Inf. Sci., Zhejiang Ocean Univ., Zhoushan, China
Abstract :
Maximum distribution reduction can obtain maximum credibility rules from an inconsistent information system. The paper analyzes various situations may be encountered when incremental objects add to an inconsistent information system, and then provides ways about how to incrementally update discernibility formula. The analysis results show that, in most cases, with the incremental object coming, discernibility formula only need incremental updating, But when incremental object causes values of maximum credibility function of two equivalence classes from different to same, because there is no inverse operation of conjunctive operation, incremental computing can not be performed. Complexity of the problem is analyzed finally.
Keywords :
Data mining; Information analysis; Information science; Information systems; Machine learning algorithms; Mathematics; Oceans; Physics; Rough sets; Set theory; Rough set; attributes reduction; inconsistent system; maximum distribution reduction;
Conference_Titel :
Networked Computing (INC), 2010 6th International Conference on
Conference_Location :
Gyeongju, Korea (South)
Print_ISBN :
978-1-4244-6986-4
Electronic_ISBN :
978-89-88678-20-6