Getting
Your Research Program Started
Seminar
notes from a talk given on September 22, 1993
by
David Whalley & John DeCatrel at Florida State University
1.
How does one identify a faculty member to work with?
- Talk
to faculty members. Find out what they are currently working on and if
you are interested. They may (or may not) want to work with you. See
what sort of research funding is available.
- Talk
to current or former students. Are the demands placed by the faculty
member one that you would enjoy or, at least, be able to handle? On a
personal level, will you be able to work with, and take instructions from,
this person?
- You
may want to make inquires at other institutions, especially at the PhD
level.
2.
How does one get started in research?
- Assume
that you have identified an area that you and your prospective advisor
will work in. Your advisor may have a specific reading list and
funded/unfunded project.
- Take
any prerequisite courses, then advanced courses in the area offered
especially, but not necessarily limited to, the advisor.
- There
may be tutorial books or collections of papers in the area published by,
e.g., ACM, IEEE, Morgan-Kaufmann.
- There
probably are collections of classic papers, and some classic books that
you should read. These are generally not state of the art, but are
invaluable in giving you a solid foundation.
- Read
current papers in conferences and journals in your new field. Attend a
conference or two if feasible.
3.
How does one come up with an idea?
- Keep
current. Know what is going on in your areas of interest. It isn’t
useful to come up with an idea that has already been investigated.
- Don’t
always accept what you read. The author will write a paper to make it
look good. If there are quantitative measurements, then check the details
(input data, test programs, etc.). Often the conclusions may be drawn
incorrectly at the end of a paper (based on opinion rather than empirical
evidence).
- Look
at the future work section of papers. Often the authors have no intention
of following up on these ideas. If they do, then they often say they are
currently working on it.
- Attend
conferences and discuss ideas with peers. Attend the panel discussions at
conferences.
- Try
to keep your own research projects current, even if you don’t see any
immediate publications from it. Ideas for papers are often generated by
encountering difficulties and overcoming these problems.
4.
What is “too large” or “too small” of a topic?
- Most
likely the topics you examine to write a thesis will be too large. A
broad problem is usually very difficult to solve. The problem may have
already been partitioned and its pieces researched by dozens of people.
One of these pieces may be suitably narrow, and have some remaining open
questions that you can work on.
- Look
at other people’s theses and get a feel for the scope of their problems
addressed. Of course, it’s better to examine “good” theses. There are
plenty of poor ones out there. Obviously theses in your area, perhaps
available via inter-library loan, serve as better references than those
outside of your area.
- The
chances that your thesis will solve some fundamental, outstanding problem
in your discipline are minuscule. It’s better to focus on a problem not
overly challenging, but terminable within a year. That doesn’t mean that
your problem has to be uninteresting. Do resist the urge to develop your
own scientific area. Incremental improvement over problems already
identified is safer.
5.
How do you pursue your idea?
- Set
up an environment to do research. A good environment will allow a
researcher to concentrate on the new idea rather than the mechanics of
setting up the experiment. Ideas can be obtained when setting up tools
and environments. Descriptions of tools and environments can also be
sources of papers.
- Start
working on an idea. You don’t have to be totally sure of it before you
pursue it. If one were sure, then it wouldn’t be research. By pursuing
the idea, one can often come up with several other new ideas. Having a
nice environment for experiments will encourage one to start the research.
- Analyze
the idea thoroughly. Pretend you are the reviewer of the paper and think
of criticisms. Try to address these criticisms (fill all the holes you
can). It is much better to thoroughly research a simple idea with limited
applications, than poorly research an idea with much greater possible
benefits.
- One
has to also understand the idea of diminishing returns. You can’t
evaluate a topic forever. If some additional analysis will produce very
limited benefits, then don’t do it.
- Look
for ideas that can be quantitatively evaluated. I am not saying that
qualitative subjects are not interesting or worthwhile. Unfortunately,
evaluation of qualitative topics is just more difficult to sell to
editors, program committees, and reviewers.