Maintenance of Knowledge in Dynamic Information Systems

Abstract

In the process of data analysis in Pawlak’s information system and in the derivation of decision algorithms from decision tables, the main computational effort is associated with the determination of the reducts. The notion of the reduct plays a major role in the definitions of other fundamental notions of the information systems theory, such as functional dependencies, redundancy etc. The purpose of the presentation is to show that role and demonstrate different ways of computing reducts. The computational complexity of the problem of reducts enumeration is, in the worst case, exponential with respect to the number of attributes. Therefore the only way to achieve its faster execution is by providing an algorithm, with a better constant factor. The algorithm presented computes reducts from the sample set of objects. The corresponding data available for the analysis in real-life applications is subject to some dynamic changes. Thus it would be useful to have a facility to support such a behaviour. The presented algorithm is based on the strategy of the maintenance of the set of reducts — providing their necessary alternations triggered by an expansion of the set of objects. The algorithm is demonstrated in a nontrivial example. The significance of the offered solution is twofold; the avoidance of the repetition of the whole (standard) procedure for the dynamically changing data, and an improvement of the execution time in comparison to the standard methods.