Parallel Session 3a

Principles of Data Mining and Knowledge Discovery

Volume 1263 of the series Lecture Notes in Computer Science pp 156-166


Generation of rules from incomplete information systems

  • Marzena KryszkiewiczAffiliated withInstitute of Computer Science, Warsaw University of Technology


In the paper we define certain and possible rules in an incomplete information system as certain/possible ones in every completion of the initial system. The careful examination of the dependencies between an incomplete system and its completions allow us to state that it is feasible to generate all certain rules and some important class of possible rules directly from the incomplete information system. Space complexity of the proposed method of rules' generation is linear with regard to the number of objects in the initial system.