Neil Walkinshaw

Research

Broadly, my research interests lie in the testing and analysis of software and adaptive infrastructure systems.

• 📚 My publications can be found here: Google Scholar | DBLP

Causal Software Engineering

My research explores techniques to support testing in traditionally difficult domains, such as complex input spaces, cyber-physical systems, and systems with long execution times. Much of my recent research has focused on the application of Causal Inference to test causal relationships between inputs and outputs in scientific software models (funded by the EPSRC CITCOM project). I have written up more details on this work in the CITCOM project page.

State Machine Inference and Analysis

I have a longstanding interest in techniques that help to understand the sequential behaviour of systems when a prior model is unavailable. Key contributions in this area include:

• Extended Finite State Machine (EFSM) inference — a technique to infer state machines with guards (Journal of Empirical Software Engineering, 2016) and full computational EFSMs with state transition functions (ICSME'16).
• LTSDiff — an algorithm to graphically compare two state machines (ACM Transactions on Software Engineering Methodology, 2013).
• Subjective Opinion State Machines — a generalisation of Probabilistic State Machines using Subjective Logic to model second-order uncertainty, improving accuracy in inferred state machines (IEEE Transactions on Software Engineering, 2023).

Second-order uncertainty in Software Engineering

Second-order uncertainty is concerned with quantifying the (un-)certainty surrounding a probability. This arises in many Software Engineering contexts - if not in most contexts where probabilistic reasoning takes place. There are various uncertainty logics out there that provide ways of reasoning about probabilities in this way. My initial forays into this area used an approach called Evidential Reasoning. I then adopted Subjective Logic, because it is more flexible and expressive. Anything you can do with a conventional probability can be extended to Subjective Logic, and thus incorporate second-order uncertainties. Key works in this area include:

• Using Evidential Reasoning to quantify second order uncertainty in Safety Arguments ISSRE'15
• Using Subjective Logic to reason about uncertainty in Empirical Software Engineering experiments EASE'20
• The work mentioned above of using Subjective Logic to reason about uncertainty in state machines (IEEE Transactions on Software Engineering, 2023).