| PLEASE NOTICE that - unless you are yourself a relic of times gone by - this page is a FOSSIL. I do not intend to take on any more research students, but I preserve the page and its descendants because others have found the material useful. I've left the text unchanged, because it's easy, but some of its implications are no longer significant |
Over the years it has become clear that there are different views in the Computer Science department ( not to mention in the world without ) as to what constitutes satisfactory performance in the various programmes of individual work which we offer to students. Some years ago, a fairly vigorous electronic mail discussion between academics in the department on the differences between our various projects petered out without reaching agreement; fairly recently, one or two students whose work I have assessed have thought my expectations unreasonable, and some unpleasantness has ensued.
I therefore thought it would be worth my trying to write down something which would give you an indication of what I expect from students engaged on such individual programmes, and of which I expect to find clear evidence in project reports and theses. I hesitated to do so, because I wasn't sure that I could be precise about just what I wanted, but concluded that something is better than nothing. That was quite a while ago now; I've tried to keep this document more or less up to date, but might well have missed bits I should have changed. Please take them as indications, not specifications; if you wish to question any points I make, by all means tell me about it, so that I can either defend my position or improve it.
The expectations work in slightly different ways in different circumstances. In assessing a project report or thesis, I expect to find evidence of the points I describe clearly presented as far as is possible. In supervising the work which leads up to the report, I expect to see evidence that work of a reasonable standard is going on fairly steadily throughout the period of the course concerned.
Obviously, those notes have to be interpreted flexibly. This particularly applies to the comments on reports - they might all have to be replaced with "an account of what went wrong, and why", interpreted more demandingly as the points increase.
A characteristic feature of projects and reading courses as compared with theses should be that projects do not require any significant originality; in all cases, it should be possible at some lowish level to prescribe what has to be done from the beginning. ( If originality happens, fair enough, but it shouldn't be expected. )
( I don't really expect to supervise any more reading courses. I include this bit because it's easier than taking it out, and just in case someone has an irresistible desire to study something in which I'm interested. If you are one such, by all means get in touch, and we can talk about it. )
I don't expect any original work, nor indeed any practical work, in a reading course. The aim is to become familiar with some area of study which isn't represented, or is inadequately represented, in the department's courses. I expect that a student will read such sources ( books, research papers, computer-retrievable information, etc. ) as are appropriate for the subject, level, and point rating, amounting to the sort of material that would reasonably have been presented in the course which we don't give.
In assessment, I would look for evidence of wide reading, and of understanding of the structure of the subject at a deeper level than superficial. I would expect to see a report ( or possibly a set of lecture notes ) in which this understanding was made clear.
A master's thesis is not just an opportunity to have fun writing lots of code. It's an academic degree, and the intention is that you should demonstrate understanding of your topic, not just programming skill. People who start a master's thesis expecting that all they have to do is to write a big programme are in for a shock.
The main emphasis of a master's thesis is on the organisation and understanding of some set task, which will almost certainly require that some sort of work - experimental, theoretical, or whatever - be carried out. I expect the student to be able to discuss the task in the context of related work ( literature review ), to analyse it and design a satisfactory research programme for its completion, and to implement the programme, and critically to analyse the results. If I'm satisfied with your programme, I don't much care whether or not it is actually completed, because things can go wrong however carefully you plan. You can still write it up satisfactorily - for example, you might demonstrate your understanding by showing that you understand what went wrong, and suggesting an alternative plan which would be more likely to succeed.
The topic need not be startlingly original, but neither should it be a repetition of other work. In particular, the student should be able to cope with original ideas, but isn't expected personally to discover anything. A new application of a well known technique, or a new way of solving some familiar problem, would be satisfactory.
The emphasis throughout is on analysing the requirements and systematically producing a solution. Knowing why whatever happened did happen is important; this applies both to the phenomena under investigation and to the conduct of the work itself. It should always be possible for the student to say "I did <X> because<Y>", or "<P> happened because<Q>".
I expect that a doctorate student will be fairly close to self-sustaining, generating his own programme of action and developing it in the light of the findings of earlier stages of the work and any relevant newly published results from elsewhere. The student must therefore actively keep up to date with developments, should certainly know more about the subject than I do, and should keep me up to date on developments of which I should be aware.
The aim of a doctorate programme is to understand something hitherto not understood, which might well be expressed in rather general terms. In this case, I do expect the student personally to break at least some of the new ground which leads to the understanding. I also expect it to be a real step forward; trivial "advances" which are obvious extensions of current knowledge won't do. It is quite likely that the immediate details of the programmes of action will be on specific topics, perhaps fairly remote from the overall aim; the student should always be able to justify each topic by showing how it is related to the aim. Ideally, the justifications should be written down before beginning work on the topic.
As a supervisor, I expect to be kept up to date with progress in the course of the work, preferably in writing. ( I'm forgetful. ) I do not expect to give instructions about what should be done next ( see above ), though I expect the opportunity of commenting on proposals for any action with large-scale consequences. I expect a student to be thoroughly acquainted with all aspects of the work done, and able to discuss it intelligently; if I ask a question, I expect either a sensible answer ( not necessarily immediately ), or an explanation of why there is no sensible answer.
I expect to do things too. I'm only writing them down once, because, except for obvious questions of scale, they're essentially the same for all sorts of individually supervised work.
I want to emphasise this point strongly, because occasionally I've supervised students who haven't kept me in touch with things, and have eventually produced reports which aren't very clear. All I can do then is wonder why they did things which seem silly, or didn't do things which seem sensible, and inevitably get a bad impression. If they'd kept in touch I'd have known more about their reasons and perhaps been able to take a more understanding view, or even suggest a better course of action.
Notice the "suggest". That's all I can do; the decisions on exactly what you do are yours. Time constraints, if no other reasons, mean that I can't be of much direct practical help.
Having said that, there is no guarantee that the plan will work. In anything with a research component, matters can turn out to be different from your expectations, and plans can turn out to be inappropriate. The plans also depend on my ( and your ) estimates of how much you can get done in the time available; if those estimates turn out to be wildly optimistic we could be in trouble, and either or both of us could be at fault. We cross these bridges if we come to them.
It might help you to bear in mind that my advice is always intended to be helpful. Yes, really. When there is some doubt, it's usually after someone has persisted in following a line of investigation which I believe to be misguided. If I've expressed my reservations several times without getting any convincing explanation in return, I begin to get rather more forthright, if not downright rude. But what else can I do ? I see the student proceeding merrily in the direction of doom, and all my warnings have been ignored; I can only shout louder.
Advice continues while you're writing your report or thesis. I want it to be good, too, and I'll read it for you and offer comments any time you want me to ( given that I do have other things to do ). I won't write it for you, but I will discuss what you're doing and how you're doing it, and make suggestions if I think fit.
I try not to begin topics unless the resources I think we need are already available or really, definitely, absolutely just about to arrive. I can't be anything like as confident about resources which suddenly turn out to be essential half way through the work. I'll still try, but prefer to try working round or choosing a different line of work; from experience, that's a better way to go given that there's limited time to get anything done.
Go to
me
( Alan Creak, in case you've forgotten );
Go to
Computer Science.
Alan Creak,
2001, February.