| Rough set theory is provided by Z.Pawlak in 1982. It is a math theory that process the non-accurate after probability theory, fuzzy theory and evidence theory. Not needing other information or previous knowledge, this theory can analyze and process the non-accurate, non-integrity and inconsistent data, then mine latent knowledge and uncover latency rule. In this dissertation, the research trends and the attribute reduction theory of RS are presented firstly. Then the existing attribute reduction methods, which consists of Pawlak algorithm; algorithms based on discernible matrix and logic operation; algorithms based on mutual information and algorithms based on information quantity, are summarized. Secondly, attribution reduction of three information systems in real world progress preliminary research, This paper shows three aspects of research results as follow:1. With respect to the imcomplete information system, an attribute reduction algorithm besed on the tolerant relation is proposed. This algorithm takes relative positive-region as the iterative criterion. In the same time, the time complexity of this algorithm is analyed, and obtained that the time complex degree of this algorithm is polynomial. In worst condition, the time complex degree is O(|C|~2|U|~2).2. With respect to the fuzzy object information system, an attribute reduction algorithm besed on the discernibility matrix is proposed. This algorithm use adequately the property of the discernibility matrix and relation between the discernibility matrix and decision reduction set, step by step to eliminate relate element in the discernibility matrix, finally obtain a minimal relative decision reduction.3. With respect to the hybrid information system of real-value attribution and symbol-value attribution, Two deferent attribute reduction algorithms are given. The first algorithm use similar discernible matrix to obtain attribute reduction; and the second algorithm use information quantity to obtain attribute reduction. |
PDF Full Text Request