IncrementalApproach for Updating Knowledge in Dynamic Data

Ronglong Zheng1,Yan Li2,3*, YanZhan2, Qiang Hua2

1JiangmenAdministration of Municipal Service and Data Management, Jiangmen, 529000,China

2College ofMathematics and Information Science,Hebei University, Baoding, 071002, China

3School of AppliedMathematics, Beijing Normal University Zhuhai, Zhuhai, 519087, China

Abstract: Dynamic data from real applications often occurs when newattributes ｏｒ objects are inserted ｏｒ old ones are removed. In the framework ofrough set theory, set approximation is one type of important knowledge whichneed to be updated from time to time for further data reduction and analysis.Some incremental methods have been proposed either for the variations inattribute set ｏｒ in the object set. In this paper, we combine the two dynamicsituations and give the principles and corresponding algorithms toincrementally updating approximations. The experimental results and analysis onfive data sets from UCI show that the incremental approach outperforms thetraditional non-incremental method especially in the dynamic situation ofremoving attributes.

Keywords: Dynamic information systems; Incremental approach; Set approximations