A metrics-based approach to the automated identification of object-oriented reusable software components

摘要

Software reuse has been a long term goal of software developers. This goal has been rather elusive, but the widespread use of the object-oriented paradigm and other innovations in software development paradigms have led to renewed interest in this topic.;This dissertation describes a new approach for the identification of reusable object-oriented legacy software components. The approach is concept-driven, knowledge-based and tool-assisted and uses heuristics to aid in the natural language understanding of comments and identifiers in object-oriented code. The approach uses a reusability quality factors hierarchy, where low-level object-oriented metrics are used to predict reusability quality factors.;The automation of this approach is provided by the Program Analysis Tool for Reuse (the PATRicia system), which is a software tool that aids in program understanding and metrics analysis. It consists of the Conceptual Hierarchy for Reuse including Semantics (CHRiS) tool, which uses a knowledge-base in the form of a conceptual-graph based semantic net in the natural language understanding of comments and identifiers, and the Metrics Analyzer tool which calculates object-oriented metrics that are then used to predict reusability metrics.;These tool-aided methods are shown to work well by identifying and qualifying reusable components in several real world graphical user interface packages. This approach demonstrates that the identification of object-oriented legacy software components can be made significantly easier and more quantifiable than is possible using earlier techniques, which will aid greatly in promoting effective software reuse.