Year: 2011
Type: dissertation
Abstract: The constant demanding, mostly from Software Maintenance professionals, so that it could be created new and more efficient methods of understanding programs, have put several challenges to Program Comprehension researchers. Nowadays, the programmers are not satisfied with the simple extraction of the program's structure, which mainly resides on the control and dataflow identification. They want to know the meaning of the program, so they can identify the real world concepts that are materialized on the source code of the program, which would provide a better and more efficient understanding of the program. To solve this problem, Program Comprehension researchers have, mainly, developed approaches that use techniques which are based on Information Retrieval Systems, like search engines. The strategy involves the retrieval of non structured information, properly ranked, answering a question the system can interpret. Considering Program Comprehension, this strategy enables the programmers to search for real world concepts, also known as Problem Domain, implemented and mapped into programming concepts, also known as Program Domain. Although its use is not consensual, source code comments have the main objective of helping understand the source code, and it has already been proven its value on the process of comprehending, through several studies. Even though the reasons for this fact have not yet been proved, some authors have defended that source code comments are an important vehicle for the inclusion of Problem Domain information and that their exploration improves and increases the comprehension process. Therefore, the presenting Master Dissertation exposes the development of a solution based on Information Retrieval algorithms, based on Information Retrieval Systems, in order to check if, in fact, the information included in comments can contribute in a decisive way to increase the efficiency of the comprehension process of a given program or software system. The results and conclusions, extracted from this work, showed that comments, properly analyzed and classified by the developed system, helped to better understand the Problem Domain concepts and their materialization on the source code. The developed solution showed itself able of meeting the challenges posed, and proved to be a usefull and efficient tool for the comprehension tasks that may emerge on the process of software maintence of a system.
Authors: José Luís, Figueiredo de Freitas
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2161093078