The Research Of Aspect Mining And Its Application In Program Comprehension

Maintenance is becoming more and more important in the process of software development, which takes a large portion of the entire life cycle of software. The efficiency of the maintenance depends a lot on the work of comprehension of the software by the developer. Program comprehension is applied in many fields such as software reverse engineering and software analysis, in which the key work is how to extract the information in software code, and then build up the mental model to support, at last, information will be organized in some forms and showed to the user.Concerns in the program design mean the partition in the software which is describable and manageable. Due to the defect of the traditional program design method, such as OOP, lots of legacy software system has the phenomena of crosscut of the non-core concerns, which leads to the code tangling and code scattering. This will reduce the modularization and comprehensionable of the software system and make the maintenance more difficulty. Aspect-oriented program is a new concept of program paradigm, which puts forward the concept of aspect in the basic of the object-oriented program which is refactored by identifying and mining the code of crosscut concerns. Aspect mining is the main job, which analysises, calculates and mines the crosscut concerns, and then build up candidate aspect, preparing for the following work of aspect refactor.After analysising the existed program comprehension tools, we find that, these tools can't support enough to the aspect-oriented program. So, this study suggests that aspect mining as a part of information extraction, apply it to the program comprehension after improving the algorithm of aspect mining. The research has done some contribution as follows:(1) This study researches the traditional aspect mining algorithm. After comparing them, we put forward a improved algorithm: a aspect mining algorithm based on Fan-in analysis and clone detection, in which the clone detection is improved on the traditional PDG algorithm. The case study proves that this algorithm has better performance on mining time and mining result than the traditional ones.(2) This study also designs and developed an aspect-oriented information extraction system of program comprehension. First, cognitive theory is researched. On the base of synthetical cognitive theory, the functional requirement is researched. This system can analyse the software in static ways, and as a part of information extraction, the result of aspect mining will also be showed to the user graphically. The analysis of the performance of the system also proves that it optimized in aspect-oriented level compared with some traditional program comprehension tools.