A spotlight on human factors research of Assoc. Prof. Tony Savarimuthu

posted in: Misc | 0

Associate Professor Tony Savarimuthu from the University of Otago’s Information Science department studies human factors as applied to artificial and real agents (e.g., robots and humans respectively). Thus, his research spans two main areas: multi-agent systems and software engineering.

In multi-agent systems he is interested in the design and development of simulation-based systems focussing on improving collaboration and cooperation of interacting agents (e.g., humans and robots). Particularly, his research efforts have focussed on creating socially aware artificial agents that are able to interact with humans and artificial agents in a team setting considering human factors such as social norms, expectations, trust and reputation. His earlier PhD work focussed on approaches for creating norm-aware systems where agents are able to infer and learn norms from other interacting agents.

Human factors are an integral part of most software projects. Thus, Tony’s research in the Software Engineering domain with colleagues has focussed on a range of human factors. To highlight a few, his works have investigated the nature of norm violations (e.g., coding convention violation and lack of commit comments). Norm violations are primarily violations of social expectations and these can be mined from large software repositories. When human expectations of software are violated, these are expressed in the form of user feedback (e.g. app reviews and bug reports). These expectation violations must be addressed to have a robust and resilient software eco-systems. One of his other interests is on studying decision-making processes employed in software projects. His works have proposed mechanisms to extract and verify such processes used in the Python community. A recent work that appears in ICSE 2021 investigates how rationale for making decisions can be inferred from large email repository data of stakeholder discussions. His research has also investigated the role of inferring personality types of Python core developers from written text, clustering developers’ outside of work activities that relate to their work, and more recently the use of abusive language in software development communities. His future work aims at elucidating and documenting a comprehensive set of human factors that are at play during software development in small, medium and large software teams, and also studying the influence of these factors in enabling success.