本文在对ＣＯＢＷＥＢ、ＣＬＡＳＳＩＴ等概念聚类系统研究的基础上，提出了一种用数值属性的聚类分划来表示数值属性的方法．这种表示的核心是基于数值属性的取值分布．对于在这种表示下数值属性聚类的相关问题及性质，本文进行了较为详细的讨论．在此基础上，本文给出了一个能统一处理数值属性和符号属性的聚类评价函数．一个基于聚类分划表示方法的概念形成学习系统ＣＦＬＳ（ｃｏｎｃｅｐｔｆｏｒｍａｔｉｏｎｌｅａｒｎｉｎｇｓｙｓｔｅｍ）已在微机上实现，并被应用于地质学领域的三叶虫分类问题．本文对ＣＦＬＳ的设计和实现进行了介绍 Based on the research of concept clustering system such as COBWEB and CLASSIT, this paper presents a method of representing numerical attributes by using the clustering division of numerical attributes. The core of this representation is the distribution of values ​​based on numeric attributes. For the related problems and properties of clustering of numerical attributes under this representation, we discuss in more detail in this paper. On this basis, this paper presents a clustering evaluation function that can deal with both numerical attributes and symbolic attributes. A concept-based learning system (CFLS) based on clustering and partitioning representation has been implemented on a microcomputer and applied to the trilobite classification problem in geology. This article describes the design and implementation of CFLS