| Collecting data is the use foundation of scientific research and data analysis. Rough set theory has become a powerful tool to deal with a complete information system. In real life, because of the conditions, technology, methods, limitations, as well as subjective factors and objective factors, missing data and cause the formation of incomplete information systems. For the case of missing data, from the semantics, there are two kinds of omission semantic and absence semantic. In this paper, the characteristic relation of the coexistence of two types of semantic, carrying out a method of attribute reduction research.First of all, this paper, based on the characteristic relationship in incomplete information system, re-defines the system and local completeness and incompleteness.Secondly, this paper, based on the characteristic relationship redefine granularity, entropy and condition entropy and distinguished matrix and give the granularity heuristic attribute reduction algorithm, information entropy attribute reduction algorithm and distinguished matrix attribute reduction algorithm. Through the algorithm complexity analysis, among the three methods, the distinguished matrix algorithm has the minimum time complexity.Tirdly, by use of variable precision rough set model idea, put forward the degree attribute reduction method based on the charactensticrelationship in the incomplete information system and give algorithm, and make the analysis of time complexity.At last, an example stated the effectiveness of the algorithm, as well as confirmed the degree attribute reduction method based on a smaller loss of information, the information system more simple.Finally, this paper understands the information systems and relationships by the generalization thinking, analyzes Pansystems relations during Panweight reduction, and defines importance of Panweights. Through the analysis of the process of reduction by use of Panderivative, panextrema and pancommunication, the Panweight reduction algorithm has been given. |