Dynamic Updating Rough Approximations in Distributed Information Systems

Rough set theory is an effective mathematical tool for processing the uncertainty and inexact data. In some real-life applications, data stores in information systems distributively which are called as Distributed Information Systems (DIS). It is hard to centralize the large-scale data in DIS for data mining tasks. Furthermore, knowledge needs updating as the attributes dynamically increase in size in DIS. In this paper, we present an incremental approach for maintaining rough approximations in DIS under attribute generalization. Firstly, a matrix-based approach is presented to compute approximations. Then, an incremental approach for updating rough approximations in DIS is proposed, which does not need to centralize data from different locations and recompute the whole data sets from scratch. Finally, a case study is provided for validating the efficiency and effectiveness of the proposed method.