The Knowledge Discovery Research Of Random Information System

Both rough set theory and evidence theory are the math tools for dealing with uncertain knowledge. Recently, it has been a vital studying direction that connecting rough set theory with evidence theory to research the knowledge discovery problem of random information systems. Under this studying background, the knowledge reduction and discovery problems are studied by means of the mass function in the evidence theory.First, the concept of rough degree in Pawlak approximate space (U, R) is raised which is defined by the belief measure and plausibility measure of the evidence theory. The character of the rough degree is discussed when the sets are crisp and fuzzy. The rough degree inequalities are also proved.Second, the knowledge reduction of the random information systems is researched. Relying on the mass function, the relations are studied between belief measure and plausibility measure of incoordinate objective random information systems and positive domain reduction and distribute reduction of incoordinate objective information systems. So we obtain the means of the positive domain reduction and distribute reduction of incoordinate objective random information systems based on the evidence theory.Finally, the concept of the β -approximate reduction of random information systems and objective random information systems are put forward. The character of the β -approximate reduction is analysised, which extends the former concept of the reduction of information systems and objective information systems.