【摘要】 案例推理(CBR)已经广泛地应用在非结构化和半结构化的决策问题中.CBR系统成功的关键因素之一在于它检索出最相似案例能力,即相似算法的精确性和有效性.然而现有的各种相似算法常常对案例本身有较强的假定限制,这些假定限制的存在在实际的应用领域中并没有得到证实.文章以管理案例推理系统研究为背景,提出层次相似算法模型用以解决属性间存在复杂相关性的案例检索算法,并将该理论应用于管理案例检索的实践中.更多还原

【Abstract】 Case_based reasoning(CBR) systems have widely been applied to s upport ill_structured decision making. Widely used symbolic retrieval functions, such as nearest_neighbor matching, assume independence of attributes and require spec ification of their importance for matching. To overcome this limitation, we prop ose a multi_leveled similarity measure model and apply it to the retrieval of ma nagement cases.更多还原