Abstract: In Data mining with Attribute-Oriented Generalization(AOG), attribute reduction and concept generalization could not part from concept hierarchy. Concept hierarchy is commonly transcendent, the generalization could not perform for new concept which did not presence in concept hierarchy. The measure of concept compatibility based on attribute theory, expresses single words in concept hierarchy with its Qualitative Criteria. The Algorithms of Dynamic Generation of Concept Hierachies for text attribute is given. The new value of attribute can be inserted direct into Concept Hierachies without regenerating Concept Hierachies when appearing the new attribute values. so that we cold make concept generalization and modifying concept hierarchy in synchronization, and this method makes that AOG algorithms gets the speciality of dynamic reduction.

Key words: attribute-oriented generalization, concept hierarchy tree, The measure of concept compatibility

摘要： 在面向属性归纳（AOG）的数据挖掘中，属性归约及概念归纳都离不开概念层次。而概念层次一般都是先验的，当出现概念层次中没有的新的概念时，无法进行归纳。以属性论为基础的概念相容测度，将概念层次单纯的文字表示转换为其定性基准的表示，给出了名称型属性的概念层次树的动态生成算法。当有新的属性值出现时，可以不用重新生成概念树，只须调用该算法，将新的属性值插入已有概念层次树即可。从面可以让概念归纳与修改概念层次同时进行，使AOG具有了动态归纳的特性。

关键词: 概念相容测度, 面向属性归纳, 概念层次树