Pointers for Improving an ASE Submission
A conference paper should convince the reader (to some level of detail)
of something interesting. That is, you must have something ("the contribution")
new or otherwise interesting to tell the reader that s/he did not know
previously. But the job doesn't end there; in addition to explaining the
contribution, you must convince the reader of the truth and (possibly)
interestingness of the contribution. Now, a relatively short conference
paper will not have the room to provide all the details, but that's okay.
A conference paper tells a story; keep it as simple as possible (but no
simpler). In general, you should try to write to an audience that is relatively
intelligent and technically savvy (though not necessarily expert in your
subfield) and somewhat skeptical (though not maliciously so).
Given these considerations, here are some tips that may help you improve
your paper.
- The Novel Contribution.
Be sure to say clearly what is new and interesting
about your work. This should be obvious in the abstract, introduction,
and conclusion. There can be many types of novel contributions: new tools,
techniques, or problem solving approaches; theoretical frameworks for
understanding a domain or task; models of processes, agents, or systems;
or new or replicated empirical studies, to name just a few. If you can't
put your finger on what is novel, there is no way a reader is going to
either, and it is quite likely to be rejected for that reason. "Novel"
means "new to the research community", not just to you.
- Evaluate, Evaluate, Evaluate.
If you are describing a new technique,
tool, or approach, please include some objective and convincing
evaluation of its usefulness, power, or performance. Just because you
thought of it doesn't make it necessarily interesting. The evaluation
should provide the foundation of your argument as to why the paper
should be accepted. Remember, toy examples are good for illustration,
but unconvincing as evaluations. Empirical studies having sample sizes
of one or two may give intuitions to the researcher, but they seldom
convince anyone. Of course, different types of contributions have
different natural argumentative support; e.g., theoretical papers may
be best supported by theorems instead of case studies. The key point
is to do more than merely present an idea: convince the reader of
its worth as well.
- Compare To Other Approaches.
Compare to related and contrasting
approaches in the field. This comparison should not be of the simple
form "XYZ[7] did ABC." It should be more like "XYZ[7] applied the
heuristic ABC approach to this problem as well, but their heuristic
is only applicable to boolean descriptions, whereas my approach can
apply to integer descriptions as well." Try both to situate your research
in the broader landscape and also to compare and contrast your approach
with previously published alternatives. The purpose of this exercise is
to clarify your work's novelty, not to belittle other people's work.
- Edit For Readability.
If English is not your native language, please try
to have a native speaker (preferably a native writer) help you edit it
for clarity. Even if English is your native language, please read
your draft and polish it several times. Ask colleagues to read it.
If it sounds awkward to you (or awkward to a colleague), it will read
even more so to a stranger. Believe suggestions of awkwardness.
- Keep To The Point.
Keep your writing focused on the novel contribution,
evaluation of that novel contribution, and comparison with others in the
same area. A good article maintains focus. For example, an article
describing a new synthesis approach within a tool set doesn't need a
complete description of the entire tool set (though pointers to other
papers doing just that are good). "Omit needless words." Know your
audience.
The OOPSLA conference has a good and more detailed
page of advice
on submitting papers that is worth looking at as well
(particularly the 1991 paper by Snyder)