Abstract：The information system with incomplete interval value was investigated to propose a new improved model. The definition of similarity was provided based on deviation degree to deal with the situations that most attribute values were intersected but not included and contained multiple continuous values. The tolerance relation was extended to establish dualvariable precision tolerance relation based on similarity and similar rate. The extension rough set model was set up for incomplete intervalvalued information system. To solve the shortage of dualvariable precision tolerance relation in dividing domain and improve approximation quality, the dualvariable precision maximal consistent classes were found out. The upper and lower approximation sets were obtained to propose the maximal classification reduction algorithm. The effective feasibility of the model and algorithm was verified by real example analysis. The results show that the proposed model is suitable for incomplete interval value date type, and can be used to deal with general incomplete interval value information system and expand the scope of rough set.
2014, Vol. 35 
