Repurposing Genetic Algorithms for Supervisor Allocations in Law Dissertations Dr Joshua Warburton, University of Leeds
Supporting Students to Understand the Law Using Revel - An Online Learning Resource Dr Val Aston, Lyndsey Davies, Maddison Lavelle, Sandra Church, and Antonia Fairbourn, Swansea University Recent research with law students at University of Salford and Swansea University suggests Revel supports students’ understanding of the law for the module, keeps them up-to-date with the reading for the module and helps them prepare for seminars confidently. Join this session where a panel of students from Swansea University will be reflecting and discussing their experience of using Revel and the differences with traditional eBooks. The objective of this work is relatively simple: to automate staff allocation for dissertation projects, so that students are assigned to the most suitable member of staff, taking into consideration the number of students that an academic can supervise and the supervision capabilities of the rest of the faculty. The problem that arises is that staff members are expected to supervise multiple students, and students will have preferred supervisors based on their chosen topic, but it is often impossible to assign all students to their preferred supervisor. This type of problem is often called the ‘College Admissions Problem’, and I am attempting to repurpose work from other disciplines (notably Computer Science and Economics) to create an algorithm that deals with this problem in an efficient and accurate manner. The primary work that I am repurposing is the ‘Near Pareto Optimal Genetic Algorithm’ developed by Sanchez-Anguix, Chalumuri, Aydogan, and Julian in 2018. In this initial piece of work, they created a piece of software that could assign students to supervisors based on the algorithm - but their work is purely focused on computing, and the code is dependant upon a subject map of computing that links topics together. The biggest challenge I’ve faced is creating a subject map for law and implementing it into a modified version of their code. Despite how time-consuming this initial task has been, this work has the potential to reduce allocation-based workloads in legal subjects in a very significant way. Of course, such an algorithm does not remove the need for academic oversight entirely, and one of the major benefits of this algorithm is that it allows for relatively simple modification of the allocation (for example, if workloads need to be rebalanced somewhat). The importance of the algorithm though, is that (on the proviso that my subject map is refined enough) it will create a near-mathematically-ideal allocation of students to supervisors. I think this is a real opportunity for law teachers to use technology in a meaningful way that improves supervision, and reduces workloads.
Made with FlippingBook HTML5