Abstract: Reduction of Attribution based on rough set theory is an important but difficult task under incomplete decision-making table. Information entropy and conditional information entropy are defined to express indispensable of attributes under incomplete decision-making table, a heuristic Algorithm based on conditional information entropy for reduction of attribution is proposed. By introducing relatively positive region, problem of redundant attribution created in the course of reduction of attribution under inconsistencies system can be resolved effectively. The complexity of this algorithm is also analyzed. Finally, an illustrative example analysis shows that this algorithm can find the minimal relative reduction for decision-making tables.

摘要： 不完备信息系统中的属性约简是粗集理论应用的难点。通过引入信息熵和条件信息熵，对信息系统中属性的必要性进行了定义，提出了一种基于条件信息熵的属性约简启发式算法。通过引入相对正域，有效地解决了不一致系统属性约简过程中产生的冗余属性问题，并分析了该算法的时间复杂度。最后，通过实例说明该算法能得到不完备决策表的最小相对约简。