last edited 6 July 2003
Comments on the Project Proposals
GeneralThe front of your mark sheet was marked by your supervisor.
It concentrates mostly on the content of your research. The back of
your mark sheet was marked by someone else (in all cases but one, by me!)
It reflects how complete and well-structured your proposal was, and
whether you met all the criteria asked for at the bottom of your assesment document, under "the moderator will be asking:".
In fact, all 6 of the questions were supposed to have been answered in the
Introduction, or at least easily traced from there (e.g. The sentence "my
complete schedule is in Section 4.2" makes it easy to find the schedule.)
Very few students did this correctly, so I went to the effort of trying
to dig answers out of the rest of your sections as well. Next year's
students won't be that lucky! But they also will get a revised Assesment
Document based on the confusions your year has had.
If you got a bad mark:
I don't know if this will cheer you up, but it's much better to find out
now that you need more work on your structure or that you didn't understand
what your supervisor wanted you to do than to find out after you submit your
project. This is only one course mark, your dissertation will have
way more impact on your overall result.
Whatever mark you got:
The point of this exercise was to prove that you have learned to do and to
write up academic research. This is not quite the same as the point of the
project, which will also be marked by how good your project is, which is
partly dependent on how challenging it is. So don't necessarily take this mark as either a promise or a threat
of things to come. Issues like project difficulty will have more impact on
your dissertation mark. Hopefully both the classes & the comments
on your proposal will help elevate the mark on your final dissertation.
Congratulations on finishing the term and good luck on your projects!
Despite my lectures, notes on Dissertation Writing, and last minute proposal advice, this was one of the most frequently neglected topics. To reiterate again:
Make sure you have written one, coherent document.
( The introductory `essay' is a chapter/section that serves as an introduction!)
Make sure each subsection has its own introduction and conclusion. A
dissertation has introduction & conclusion chapters; a chapter has introductory
and concluding paragraphs; a paragraph has introductory and concluding sentences.
Fill In (Mark 2) One of the main reasons you need to know what your
thesis is is to determine what you should spend time writing about. A
good scientific document is concise as well as informative. You should only
be writing about things that are directly relevant to your thesis and advance
You shouldn't spend a great deal of time writing on "general truths" of software
engineering. It's OK to reference & briefly outline a technique you've
chosen, particularly if this helps provide / explain the structure of your
document or approach, but don't just reiterate information tangential to
your document. It's much better to have a short dissertation / document than one with irrelevant or overly pedantic information.
The point of the proposal is to get started doing research. Ideally this
will be a lot of literature research: papers and reports on similar projects,
on tools you might use, on the industry or problem you are trying to address
or assist. There can also be tool research, where you begin doing comparisons
and prototypes. You should also get as far as you can on proposing a structure
for your software: the basic architecture and representations (again, prototyping
might help here.) Writing all of these things will help you order and advance your project,
and who knows, might get you some useful feedback from your supervisor
& / or course instructor.
There is really no reason to try to fill in pages with information you learned
in your software engineering course or with protracted quotes.
use one of the two allowable reference styles (see the library handout).
I think it would be best if everyone used Harvard/Chicago style e.g. (Bryson
2003), not . This is both easiest to read (since you don't have to flip
to the back to see what's being referenced - often the author & year
is enough to recognize a paper) and easiest to write, since the tags don't
change when you add more references (not that this is a problem if you are
The only reason numeric references exist is to save space in journals. In
really tight journals like nature (where you only get 2 pages) you have superscript
or subscript numerals. Dissertations are not like this.
Regardless of how you reference your articles, a dissertation bibliography should be in alphabetic order, not order of appearance.
The Infamous Mark 6: Introductory maps to the rest of document This
isn't the entire introduction, it's just the bit of the introduction that
tells you what's in the rest of the document. This was in the requirements,
but many people didn't do it. Marking: if absent, F. If vaguely described
what's in the rest of the document, LP. If you briefly said what was in
what chapter in one paragraph G. Distinctions went to concise sections well
integrated into the argument structure that fully and clearly explained the
dissertation's structure and where things could be found.
Every entry must have:
- an author (though the author may just be an institution e.g. RedHat,
Correll, UK Home Office. But look carefully for the real author before using
an institution.), and
URLs are a *Last Resort*.
If a document came out in any other format, you should report that
instead! A dissertation is an archival document; people may refer to it
years or even decades from now. A URL may be changed tomorrow. So make
sure to give enough information so a reader in the future can read the same
document you have.
- a year (if it's a web page, look for the year of modification, otherwise put this year.)
- how it was published (e.g. journal name, book name) or the words `unpublished'
- Technical Reports and dissertations need both the author AND the institution!
The fact that it is a TR or Dissertation means that the reliability of the
information is determined by the reliability of the institution, so this
is just as important as saying what journal an article was published in.
When to reference
You should reference *everything*
you assert, that is, anything you don't prove by reason or that isn't a report
on your own experience / work. Look at any scientific paper, you'll see
every assertion has references after it. In scientific papers, a paper is
referenced *every time* information is used from it, not just once. Examples:
- Emotions have become a popular area of AI research [about 4 to
16 refs, or maybe just a couple review articles if major examples will be
explained in coming sentences in the same paragraph].
- The best way to eliminate shadows is to use a light [at least one reference, probably 2 or 3.]
The Literature Review
A lit review doesn't have to be just a liturgy of other people's projects.
In the best case it is a good explanation of a field or concept where each
assertion or explanation is properly referenced. Normally, everyone important
in a field would get referenced in such a lit review, since the way you become
important is by making points in your paper that reviews subsequently need
& want to mention. Of course if there are particular articles or books
that will have a great deal of influence on your dissertation it may well
be worth reviewing them individually in their own section.
Anyone doing iterative development should have testing start with or even
BEFORE implementation (read eXtreme programming...) It's also a very good
idea to save early versions of your program so that you can test your late
version against the early ones. It's much easier to show that feature X
makes program B better than program A than to show that feature X is "good",
because "good" tends to be a) ill-defined and b) difficult to achieve in
page author: Joanna Bryson