From tim@ed.aifh Tue May  7 10:08:48 1991
Received: from edai by aipna.ed.ac.uk; Tue, 7 May 91 10:06:55 GMT
Received: from beamish.aifh.ed.ac.uk by aifh.ed.ac.uk; Tue, 7 May 91 10:02:54 BST
From: tim@ed.aifh
Date: Tue, 7 May 91 10:04:15 BST
Message-Id: <6066.9105070904@beamish.aifh.ed.ac.uk>
To: acstaff@ed.edai, compstaff@ed.edai, techstaff@ed.edai, restaff@ed.edai
Cc: jeanb@ed.aifh
Subject: MSc project supervision ...
Status: RO


To: Teaching, Research, and Technical Staff in DAI

Below is a short information note about supervising MSc projects and
dissertations.  It is intended to be read by both fist and second
supervisors.  If you have any questions or comments arising from it, please
let me know.

Tim Smithers
MSc Projects Organiser

------------------------------------------------------------------------------


MSc in Knowledge Based Systems

Guidelines for Supervisors of MSc/Diploma Projects and Dissertations


1.0  Introduction

This note is based upon material previously prepared for distribution to MSc
project students by Grame Ritchie, Peter Ross, Bob Fisher, and myself.  It
has been modified and extended to act as a guide to supervising MSc
students.


2.0 General Perspective

The MSc project and thesis do not have to be original.  However, it should
be practical rather than purely a discussion of theory or a survey of other
people's work.  The first two terms of the MSc course are intended to give                    
students the basic concepts of the field, and the project work is intended
to enable them to develop some practical skills to complement this.  The
ability to build non-trivial AI systems is one of these skills, but not the
only one.  These are some of the others:

 * The ability to build an AI system in such a way that others can
   understand it.  This calls for good design, construction, 
   documentation, and reporting.

 * The ability to test ideas in the form of an experimental or prototype
   system, to assess whether it works or not, and to be able to put forward
   evidence and explanations that others can accept for its success or
   failure.

 * The ability to sustain a significant enterprise and make progress with
   it, judging what constitutes progress, what is preventing progress, and
   when it is sensible to amend the goals.

Given these general points about the nature and purpose of MSc projects,
supervisors should see that students first construct a clear plan of action.
This should be used to both judge progress and when and how project
objectives and goals need to be modified.  It should also be used for
reporting against.  This is best done by requiring the student to produce a
written report of work done, goals achieved, and problems experienced prior
to each supervision session.  Agreed modifications and immediate project
activities should also be documented by the student after each supervision
session.

The final dissertation should show how a student has gone about developing
such skills.  It should describe the goals of the project, and any changes
made to them in the light of experience, and why.  It should present
sufficient context, in terms of work done in Edinburgh or elsewhere, to show
why the goals are interesting--students should, however, assume that the
reader is knowledgeable, although perhaps with a different perspective on
the subject.  This means that students should not regurgitate lecture
material, nor spend much space on defining well-known terminology and ideas.
The overall flavour of the thesis should be a record of sustained and
well-applied effort, describing the development of the ideas and their
expression in some kind of practical system.  A common mistake made by
students is to believe that a description of the structure of a system
fulfills this; it does not!  Rather, there should be an account of how parts
of the system relate to ideas about attaining the project goals, and why,
and how the ideas were or could be tested.  It is unwise to include large
chunks of program code in the text, for example, since nearly all programs
consist largely of uninteresting scaffolding that supports the key
fragments.  Students should be encouraged to show that they
appreciate the distinction.  The key issues are:

 a) balance,
 b) good judgement,
 c) sustained and coherent effort, 
 d) relevance to others understanding and efforts, if not active research.

All this is not to suggest that projects should achieve all their originally
stated goals.  This is not necessarily so.  It is equally worthwhile for
students to fail to achieve what originally seemed viable, provided that the
final dissertation explains why the aims had seemed feasible and how the
problem was explored to discover that they were not.  Such an exploration if
thoroughly done and well analysed can be much more valuable as a
dissertation than a straightforward record of success.  The five months
allotted for the project is not long, and it is rare to be able to produce a
complete and self-contained piece of useful work in that time.  It is,
therefore, a good idea for students to include in their dissertation a
discussion of how they would have developed the work given more time.  This
can also mean describing how they would have set about identifying the
problems that they finally hit, or how they might have solved them if they
had already identified them, or how they would have changed the aims of the
work to avoid them.


3.0  Timing

Five months is remarkably short.  It is therefore vital that students get
off to an effective start straight away.  This means that supervisors need
to ensure that students understand what they are trying to do and have clear
activities and timescales set out and agreed by you, at least in the early
stages.  Later the more able students will be able to do more of this
themselves, but _do_not_ assume that they will!  Some students will need
close supervision, sometimes of quite a detailed nature, all the way to the
end.  Never assume students are doing ok, always establish that they are.
If necessary, get them to produce written reports and/or demonstrations
during the project.  These will be useful to them when they come to write up
any way.

There is an MSc project review meeting held in early July (Tuesday 2nd July
this year) for which you will be asked to produce a written progress report
for all the students you supervise.  For this to be effective it is
important that students are well under way by then and making good progress.
Otherwise, identifying problem cases and prescribing remedial action becomes
more difficult.

It is up to you how often you see your students, but once every one or two
weeks is the norm.  Fixing regular times in the diary is a good thing to do
as it reduces the overhead of remembering to contact the student, fix a
date, remind them to produce a report for the meeting, etc.  This is also
good for the students since they will have to get used to regular formal
reporting practices whatever they do after they complete the MSc.



4.0 Final Dissertation

The final dissertation should not be very long.  About 70 normal single
spaced A4 pages is usually about right for the main text (excluding
appendices).  The department limits MSc dissertations to less than 20,000
words and 15,000 has been the average.  (This limit is for the main text and
does not include appendices and tables, etc.)

Students should begin to write up at the beginning of August in time for
final submission two weeks before the beginning of the new term in October.
This means by the end of the third week of September.  Writing up should
start with the student producing a dissertation outline.  This should
identify chapters and major sections and contain paragraphs summarising the
contents and purpose of each chapter and section.  Once this has been agreed
by you, a timetable should be constructed which identifies the order in
which chapters are to be written, the time allocated to each one, the dates
by which draft chapters will be ready for review by you and the second
supervisor, and, of course, the date by which this will all be put together
as a final draft, which should also be reviewed by you prior to final
submission.

Unless a student has a very good excuse (e.g. serious illness, last week
illness, death in the family, etc.)  their is little chance of getting any
extension on the final submission date.  (Bad planning, too many
distractions,too hard a problem, are not excuses.)  This is a Faculty
decision and the Department has no say in extensions at all.  If you feel
there are grounds for an extension, you should discuss the matter with the
MSc project organizer as soon as possible.  They will have to write letters
to Faculty requesting the extension.  Do not wait until the last moment.


Tim Smithers
Revised May, 1991

