Software sustainability evaluation has become an essential component of software engineering (SE) owing to sustainability considerations that must be incorporated into software development. Several studies have been performed to address the issues associated with sustainability concerns in the SE process. However, current practices extensively rely on participant experiences to evaluate sustainability achievement. Moreover, there exist limited quantifiable methods for supporting software sustainability evaluation. Our primary objective is to present a methodology that can assist software engineers in evaluating a software system based on well-defined sustainability metrics and measurements. We propose a novel approach that combines machine learning (ML) and software analysis methods. To simplify the application of the proposed approach, we present a semi-automated tool that supports engineers in assessing the sustainability achievement of a software system. The results of our study demonstrate that the proposed approach determines sustainability criteria and defines sustainability achievement in terms of a traceable matrix. Our theoretical evaluation and empirical study demonstrate that the proposed support tool can help engineers identify sustainability limitations in a particular feature of a software system. Our semi-automated tool can identify features that must be revised to enhance sustainability achievement.

中文翻译：

---

评估软件可持续性的软件分析方法

由于必须将可持续性考虑纳入软件开发，软件可持续性评估已成为软件工程 (SE) 的重要组成部分。已经进行了几项研究来解决与 SE 过程中的可持续性问题相关的问题。然而，目前的做法广泛依赖于参与者的经验来评估可持续性成就。此外，支持软件可持续性评估的可量化方法有限。我们的主要目标是提出一种方法，可以帮助软件工程师根据明确定义的可持续性指标和测量来评估软件系统。我们提出了一种结合机器学习 (ML) 和软件分析方法的新方法。为了简化所提议方法的应用，我们提出了一种半自动化工具，支持工程师评估软件系统的可持续性成就。我们的研究结果表明，所提出的方法确定了可持续性标准，并根据可追溯矩阵定义了可持续性成就。我们的理论评估和实证研究表明，所提出的支持工具可以帮助工程师识别软件系统特定功能的可持续性限制。我们的半自动化工具可以识别必须修改以增强可持续性成就的功能。我们的研究结果表明，所提出的方法确定了可持续性标准，并根据可追溯矩阵定义了可持续性成就。我们的理论评估和实证研究表明，所提出的支持工具可以帮助工程师识别软件系统特定功能的可持续性限制。我们的半自动化工具可以识别必须修改以增强可持续性成就的功能。我们的研究结果表明，所提出的方法确定了可持续性标准，并根据可追溯矩阵定义了可持续性成就。我们的理论评估和实证研究表明，所提出的支持工具可以帮助工程师识别软件系统特定功能的可持续性限制。我们的半自动化工具可以识别必须修改以增强可持续性成就的功能。