Implementação de técnicas semânticas de mineração para identificar indícios de interesses transversais em softwares orientados a objetos

Abstract

Maintenance activity is performed aiming modify the software structure ensuring their integrity. And this task becomes easier when we have a full understanding of several system components behavior. Aiming to gain this understanding, researchs emerged suggesting approaches to identify and separate codes of a specific functionality which are entangled or scattered through various application modules, called crosscutting concerns. Considering these facts, are presented in this paper the behavior of certain techniques for crosscutting concerns indications detection and a comparision criteria considering results granularity, user involvement, symptoms of existence of crosscutting concerns and informations types required to technique computation. In addition, is discussed in this paper the development and comparison of three computational tools as plug-ins for Eclipse environment aiming to detect crosscutting concerns indications in Object-Oriented Software in Java. The execution and assessment of theses plug-ins were performed in three real software systems and the results are presented considering criteria as covering and relation between execution time and the software size.