Building software quality models for computer games

电脑游戏产业(CG)是软件产业的重要组成部分。游戏是一个亿万富翁的市场，每年为加拿大国内生产总值增加37亿加元，支持超过4万个工作岗位[2]。然而，软件工程(SE)社区很少研究计算机游戏，而且缺乏关于CG的研究，因为有一种观点认为CG更像是艺术而不是工程。我的研究项目专注于建模、测试和实验，为计算机游戏软件工程(SEGA)的发展做出了贡献。该计划的长期目标是制定经验模型和方法，以解决CG工程学的不足。它的活动将产生贡献，拓宽CG工程的理论知识，同时也出于对使用的考虑。在短期内，本提案的目标是培养两名博士生和四名硕士研究生，并分为以下三个目标。 1.为电脑游戏测试建立模型和新技术。第一个目标侧重于在DevOps环境中应用自动转换(例如，测试放大)，目标是及早检测回归错误。此外，使用机器学习技术(强化学习)，这一目标将研究在CG项目中自动生成和测试大量系统配置的新方法，以降低手动测试的成本。 2.建立计算机游戏调试的模型和新技术。与测试一起，调试是发现故障的一个必不可少的过程。以前的结果讨论了解决软件开发中的调试挑战的人群方法。因此，第二个目标侧重于调查计算机游戏的调试。因此，使用协作方法(如群调试)调试计算机游戏的努力是一个值得探索的研究机会。 3.构建电脑游戏的实证理论和质量模型。对游戏机测试和调试现象的深入理解为探索CG的新理论和质量模型提供了机会。因此，这一目标集中在与CG行业(如育碧)密切合作的研究，应用于以下领域：(1)SEGA中的人为因素；(2)数据驱动和机器学习SE以改进计算机游戏中的软件质量实践；(3)构建软件质量模型以支持CG问题。 因此，我的项目为电脑游戏行业的质量提升和维护成本的降低做出了贡献。

Computer games industry (CG) is an essential part of the software industry. Games are a multi-billionaire market, adding CD$ 3.7 billion to Canada's GDP annually, supporting more than 40,000 jobs [2]. However, the software engineering (SE) community rarely studies computer games, and there is a lack of research about CG because there is a vision CG is more art than engineering. My research program contributes to the development of software engineering for computer games (SEGA) by concentrating on modelling, testing, and experimentation. The long-term objective of this program is the formulation of empirical models and approaches to address the lacks of CG engineering. Its activities will produce contributions that widen theoretical knowledge of CG engineering while being also motivated by considerations of use. On the short-term, the present proposal aims at training two doctoral students and four master students and is divided into three objectives as follows. 1. Build models and new techniques for computer game testing. The first objective focuses on applying automatic transformations (e.g., test amplification) in the context of DevOps and targets the early detection of regression bugs. Further, using machine learning techniques (reinforcement learning), this objective shall investigate new approaches to automatically generate and test large numbers of system configurations in CG projects to reduce the costs of manual testing. 2. Build models and new techniques for computer game debugging. Together with testing, debugging is an essential process to find faults. Previous results have discussed crowd approaches to address debugging challenges in software development. Thus, the second objective focuses on investigating debugging for computer games. Thus, the effort to debug computer games using a collaborative approach (as Swarm Debugging) is a research opportunity to explore. 3. Build empirical theories and quality models for computer games. A deep comprehension on testing and debugging phenomena for game computers opens an opportunity to explore new theories and quality models fro CG. Thus, this objective focuses on researching in close cooperation with CG industry (like Ubisoft), applying in the following areas: (1) human factors in SEGA; (2) data-driven and machine learning SE to improve software quality practices in computer games; (3) build software quality models to support CG issues. Accordingly, my program contributes to advancing of quality and reducing maintenance costs to computer games industry.