4.4 Dynamic Matchmaking

"Free-for-all" allocation mechanisms which involve students (and/or staff) selecting projects themselves are always time-consuming and often unsatisfactory for a substantial proportion of those involved.

---ooOoo---

This bundle provides a mechanism for establishing student-supervisor pairings in an environment of large numbers of staff and students?probably unknown to each other?in which there is a possible skills-and-interest mis-match.

The way it works is to partition the topic areas that projects occupy as uniformly as possible (for example, AI, graphics, theory ?) - this partition is of necessity coloured by local supply and demand, but may be amended each year on the basis of demand experienced the year before, and adjustments to the supervisors' profile. In most cases, the assignment of supervisors to areas is quite easy to determine, although it is probable that some will be more adaptable than others.

Students are invited to nominate three (say) topic areas in which they are prepared to work - these would normally be prioritised. Where pre-specified projects exist or have been negotiated, they will normally belong to one of these categories and a preference for it may be indicated by the student.

At this stage the complexity of the allocation problem is much reduced and it is a feasible task to construct by hand a supervisor-student pairing by matching on preferred topic areas. It is important that this is done by a single individual who has an overview of all the issues and personnel involved, and in whom both students and supervisors have trust. It is possible that software could ease this task, but it is likely that knowledge of local personalities (either student or staff) would always be needed to fine-tune the allocation. The allocation may then be used as the basis of negotiation toward a full project specification - a nice feature is that the final project may potentially be nothing to do with the originally nominated topic areas, provided supervisor and student agree.

Given that there are staff to cover enough topics (not a problem in a moderate-sized community) this mechanism can remove the "beauty competition" aspect of allocation by making it difficult for students to pursue individual staff as supervisors.

It works better if the match of staff and student preferences coincides well with the topic areas defined. There is an obvious element of serendipity in this, but some preparation can be done by ensuring that all reference to the project in earlier years is in the vocabulary of the topic split preferred.

It doesn't work if significant numbers of students are unwilling or unable to complete forms (in which case their allocation has to be random), or if significant numbers of supervisors are very restrictive about the types of projects with which they are prepared to become involved. It is also necessary that all parties abide by the allocator's decisions.

---ooOoo---

So: ensure allocation is under the control of a single individual, and give them a mechanism to make the task manageable.