Back to search

Πανεπιστήμιο Μακεδονίας

Graph-based software evolution analysis of object-oriented systems

Abstract

dc:description

Contemporary software systems evolve over a large number of versions due to the need for continuous adaptive and corrective maintenance. The duration and cost of this process call for software systems that can be extended or modified in an effortless and rapid way. To this end, the maintainability of such systems can be enhanced by taking advantage of software history and analyzing the evolution of the source code and other artifacts stored in version control systems. In this context, inspired by applications of networks in other domains, this thesis proposes graph-based tools and techniques for different aspects of software lifecycle that facilitate the evolutionary analysis of object-oriented systems. A fundamental process that determines not only the structure of a software system but also the future maintainability, is the distribution of feature implementation code to different software modules, also known as feature scattering. Towards the better understanding of this phenomenon, a set of techniques for the analysis of the evolution of feature scattering, based on the classes and methods involved in the implementation of high-level, distinct and observable pieces of functionality is introduced. The proposed analyses have been applied on several versions of four open-source projects. Based on the results, the applied techniques appear to be promising since they allow software stakeholders to assess the evolution of feature scattering in various ways, thus gaining insight into the associated implications. Furthermore, in this thesis an attempt to develop a prediction model capable of forecasting trends in the evolution of networks representing Java systems has been made. The proposed model incorporates findings regarding the growth patterns of software networks, such as the conformance to the preferential attachment model, the duplication model, the effect of node age, and the number of node and edge removals. Moreover, by acknowledging the importance of domain specific characteristics, the model has been enhanced by rules specific to object-oriented design. Evaluation against ten open-source projects showed that the proposed forecasting approach can provide sufficient insight to the future evolution trends. Software maintenance can benefit from the application of such models by focusing on parts of the network whose properties tend to deteriorate. Finally, by acknowledging the lack of a tool for effortless software evolution analysis, a platform for the facilitation of software engineering research has been developed. The tool under the name ‘SEAgle’ is a web-based platform that enables users to analyze any git open source repository that is freely available on the internet. SEAgle calculates a multitude of metrics of three different categories and publishes the results in a free online web page and renders them available through well documented REST web services.

Degree

thesis:*
Grantor dc:publisher
Πανεπιστήμιο Μακεδονίας
Year dc:date
2016

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Chaiκalis, Theodore
  • Χαϊκάλης, Θεόδωρος

Subjects

dc:subject × 38

Rights

dc:rights
Statement dc:rights
  • BY_NC_SA
Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/43289

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Chaiκalis, Theodore; Χαϊκάλης, Θεόδωρος. Graph-based software evolution analysis of object-oriented systems. Πανεπιστήμιο Μακεδονίας, 2016. http://hdl.handle.net/10442/hedi/43289