【摘要】 形式化概念分析已被证明是一种支持模块重组的有效工具.然而,在使用概念分析进行模块重组时,仍存在一些问题有待进一步解决.本文提出一种自动化的遗留软件重组方法,用于从识别的模块化方案中获取一种最佳解决方案.通过在概念分析过程中引入概念抽象、哑概念和概念相似性度量等新的特性,有助于解决现有方法的不足.概念抽象用于表示候选模块,它消除了对概念外延的约束;概念相似性度量用于解决概念间的重叠问题;哑概念有助于减少人工干预,提高模块重组的自动化程度.本文给出算法的具体实现,并使用实验对方法的有效性进行了验证.更多还原

【Abstract】 Although formal concept analysis(FCA) has been proved to be an effective tool supporting module restructuring,some problems have yet to be resolved.This paper presents an approach focused on automated support for restructuring modules existing in legacy software.Three new features,concept abstraction,dummy concepts and similarity measure are explored to identify candidates of modules and select a best one among these candidates during concept analysis.Concept abstraction is introduced to represent a potential module,removing the over-restrictive constraints on concept extents.A new similarity measure is defined to classify concepts if concepts overlap in a lattice.It is studied specifically from information entropy perspective.The dummy concept is proposed to minimize human intervention and increase the degree of automation of module restructuring.A concrete algorithm implementation is proposed and several experiments are conducted to validate the effectiveness of the approach.更多还原