Chemistry 版 (精华区)
发信人: loafer (快乐的化身), 信区: Chemistry
标 题: [转载]Hints_On_Writing_Technical_Papers.pdf
发信站: 哈工大紫丁香 (Sun Aug 24 22:00:09 2003)
关于写论文和作ppt的好文章!
推荐!!
134 IEEE TRANSACTIONS ON EDUCATION, VOL. 42, NO. 2, MAY 1999
Hints on Writing Technical Papers and Making Presentations
Victor O. K. Li, Fellow, IEEE
Abstract:
this paper is an attempt to give some guidelines
on how to write a good technical paper and to make good
presentations, important skills for a successful career in research
and teaching.
Index Terms:
technical presentation, technical writing, thesis
presentation.
I. INTRODUCTION
I have been involved in a career in research and teaching for
17 years. During this period, I have had the opportunity
to work very closely with some very bright Ph.D. students.1
One of the major problems that they ran into, as I myself ran
into earlier in my career, is how to communicate effectively.
Over the years, my own techniques have improved. This
was achieved by reading books on the subject, talking to
colleagues, and above all, observing and studying the techniques
of those considered good writers and good speakers in
the research community. In this paper, I will summarize my
findings.
In the next section, the structure of a technical paper is
described. Section III contains hints on making good presentations,
and Section IV, the conclusion. Hints on the thesis
defense are included in Appendix A. Some common writing
pitfalls are included in Appendix B.
II. STRUCTURE OF A TECHNICAL PAPER
A technical paper should be clear and concise. The goal is
to convey ideas and results to the readers in the least possible
time and space. The notations and format should be consistent
throughout the paper. A paper usually consists of the following
components:
1) Title-It should be concise and to the point. For example,
some publications limit the title to less than ten
words.
2) Abstract-A summary of the paper, including a brief
description of the problem, the solution, and conclusions.
Do not cite references in the abstract.
Manuscript received May 8, 1995; revised December 14, 1998.
The author is with the Department of Electrical and Electronic Engineering,
The University of Hong Kong, Hong Kong, China.
Publisher Item Identifier S 0018-9359(99)03957-6.
1The primary audience of this paper was originally my Ph.D. students. I
am glad to report that the time I have spent in preparing this paper has been
more than offset by the savings in time I would have had to spend correcting
their writing. A colleague suggested that I submit this paper for publication
since other engineering educators may find it useful.
3) Keywords hey should be selected such that a computerized
search will be facilitated.
4) Introduction his should contain the background of the
problem, why it is important, and what others have done
to solve this problem. All related existing work should
be properly described and referenced. The proposed
solution should be briefly described, with explanations
of how it is different from, and superior to, existing
solutions. The last paragraph should be a summary of
what will be described in each subsequent section of the
paper.
5) System Model he proposed model is described. There
will invariably be assumptions made. State the model
assumptions clearly. Do the assumptions make sense?
Sometimes it will be necessary to introduce assumptions
to make the problem mathematically tractable, but they
should at least reflect some real-world situations. Thus
while the proposed assumptions may not hold in general,
there should at least be some instances where they will
hold and hence the model and conclusions drawn will
apply. Use figures to help explain the model.
6) Numerical results-Based on the model, numerical results
will be generated. These results should be presented
in such a way as to facilitate the readers?understanding.
Usually, they will be presented in the form of figures or
tables. The parameter values chosen should make sense.
They should preferably be taken from systems in the
real world. If that is not possible, they may correspond
to those values for which published results are available
so that one can compare these results with existing ones.
All the results should be interpreted. Try to explain why
the curves look the way they do. Is it because of the
assumptions, or because the system behaves that way?
In most cases, a simulation model is required to validate
the system model with the assumptions. In that case,
it is important to not only show the average values of
the simulation results, but also confidence intervals. A
number of books on modeling techniques will show how
confidence intervals may be obtained. In addition, it is
very important to explain what is being simulated. I
once attended a thesis defense in which the candidate
claimed that his analytical results matched the simulation
results perfectly. I then found that he had basically
simulated the analytical model, using the same assumptions!
Details on the simulation time, the computer,
and the language used in the simulation should also be
included.
0018?359/99$10.00 ?1999 IEEE
LI: HINTS ON WRITING TECHNICAL PAPERS 135
7) Conclusions his summarizes what have been done
and concluded based on the results. A description of
future research should also be included.
8) References his should contain a list of papers referred
to in the paper. If there is a choice, use a reference
which is more readily available, i.e., if an author has
published a conference version and a journal version of
the paper, refer to the journal version. Research reports,
internal memos, private correspondences, and preprints
are usually hard to access and should be avoided as
much as possible. By the way, journal editors tend to
pick reviewers from the authors of the references cited
in the submission.
9) Appendix hose materials which are deemed inessential
to the understanding of the paper, but included for
the sake of completeness. Sometimes, detailed mathematical
proofs are put in the appendix to make the paper
more readable.
10) Figures he figures may be placed immediately after
they are referred to in the text, or placed at the end
of the paper. Each figure should be readable without
relying on the accompanying description in the text.
Thus, all symbols used in the figure should be explained
in the figure legend. In addition, do not make the figures
and legends too small. Some figures may be reduced by
the publisher before they are printed, and one should
ensure that the figures are still legible after reduction.
III. PRESENTATIONS
Good presentation skill is another prerequisite of a successful
researcher. Good ideas will not be recognized unless they
are effectively conveyed to others. The presentation should be
rehearsed. This will help determine how many slides should
be included in the presentation. A good rule of thumb is one
slide per minute of presentation, although that depends on
the individual speaker and the contents of the slide. Friends
can be invited to the rehearsal, to ask questions, and to give
suggestions for improvements. Some other points to note are:
1) Visit the venue of the presentation before the talk to
get familiar with the layout of the room, and with the
presentation equipment.
2) Do not dive into the viewgraphs immediately. Establish
rapport with the audience by speaking to them for a
couple of minutes before turning on the projector. What
does one talk about? One can summarize the results;
explain how this talk relates to other talks in the session;
or how it relates to the keynote speech of the conference;
or even tell a story of how one got interested in this
particular research topic, etc. Of course, this cuts into
valuable presentation time, and should not be overdone.
This will get the audience to focus on the speaker, rather
than on the screen.
3) Maintain eye contact with the audience throughout the
talk. Talk to the audience, not to the viewgraphs. Address
different parts of the room as the talk progresses. This
means that the room should remain lit. Most overheads
will still be readable with the room lit. It is a bad idea
to speak in a dark room in which the only thing visible
is the screen. The contact with the audience will be lost.
4) Make sure the visual aids are readable. There are now
a large number of presentation tools available, ranging
from handwritten transparencies to multimedia presentations.
The most popular one is still transparencies
projected onto a screen using an overhead projector.
To determine if a transparency will be readable, the
following test can be used. Put the transparency on the
floor. It should be readable while the reader is standing.
Using a bit of color to highlight important points is useful,
although color transparencies are more expensive.
With the increasing popularity of video projectors, it
is expected that most presentations in the future will
be made using a notebook PC connected to a video
projector. It will be a good idea, however, to bring
transparencies as a backup.
5) Do not put too many ideas on the same transparency.
The audience should be focusing on what the speaker
has to say, rather than desperately trying to read the
transparency.
6) Everything on the slide should be explained.
7) The presentation does not have to follow the paper
exactly. In a conference, 25 min will normally be
allocated to each speaker, with 20 min for presentation
and five minutes for questions and answers, and there
is no way to include all the details of the paper. The
goal should be to explain the importance of the work,
the key ideas of the solution, and how it is different, and
hopefully, better than existing solutions. If the audience
is interested, the paper is available in the conference
proceedings. I have attended a presentation in which the
speaker just made transparencies of the paper and read
the transparencies. Needless to say, this is unacceptable.
8) On the other hand, the presentation should not deviate
too much from the paper either. I once attended a presentation
in which the speaker said that the paper in the
proceedings described old results, and then proceeded to
deliver an altogether different paper. While this may be
acceptable for informal workshops, it is not appropriate
for conferences in which all of the submissions have
been formally reviewed.
9) Do not put too much mathematics on the slides. It is usually
difficult to follow detailed mathematical derivations
during a 20-min presentation. Just enough mathematics
should be presented to bring the key points across. The
focus of the talk, in general, should be on the results.
Use figures (plots) to bring the points across.
10) There will usually be questions at the end of the
presentation. In fact, a good session chair will usually
prepare a couple of questions in advance just in case
there are no questions from the audience. Some of
these questions will hopefully have already been asked
during the rehearsal, and should be handled very well.
If there is difficulty with a particular question, do not
be overly defensive. While there may be showoffs
who deliberately make very critical remarks just to
show how good they are, in general most people are
136 IEEE TRANSACTIONS ON EDUCATION, VOL. 42, NO. 2, MAY 1999
just trying to be helpful, and perhaps have not really
understood some key points of the presentation. If
forced into a corner, and the session chair does not
come to the rescue, one can escape by suggesting that,
due to the lack of time, the discussion will be continued
during the coffee break. Alternatively, a particularly
difficult question posed by the audience can be turned
into an idea for future work. By the way, it is always a
good idea to repeat the question so the audience knows
what is being asked. This will also ensure the question
is understood.
11) It is a good idea to have a slide entitled 揅ontributions,
攅specially in an interview presentation. I have been to
many interview talks at the end of which I was not
sure what the speaker抯 own contributions were. Never
leave an interview committee in doubt about the nature
of the contributions.
IV. CONCLUSIONS
Some thoughts on how to write good technical papers and
make good presentations are included in this paper. Hopefully,
this will help the reader communicate ideas and results to the
research community more effectively.
APPENDIX A
THESIS DEFENSE
During the defense, the committee is going to look for:
1) whether the candidate has a very good understanding of
the problem. That means one should be conversant in the
relevant literature, and should explain how the proposed
approach is different and better than existing work. In
addition, be ready to discuss how practical the work is,
i.e., is it just an academic exercise or can one actually
use it in the real world? Are the assumptions realistic?
2) the contribution to the research community. This should
be described in the abstract of the thesis, and repeated
in the conclusions. There should be at least one slide
entitled 揅ontributions?in the defense presentation.
3) whether the candidate has taken the time to digest the
results generated. That is, whenever results are shown,
such as a plot, or a table, be prepared to interpret
the results. An intuitive explanation of why the results
look a particular way is especially helpful. If there are
simulation results, explain what is being accomplished
with the simulation, e.g., to justify a certain assumption
in the analytical model. Be sure to have explanation for
results which look strange, such as a curve which is not
smooth, or a table with sudden jumps in values.
4) suggested future work. List a few possible directions.
Describe possible approaches to these problems, i.e.,
demonstrate that these problems have been thought
through.
APPENDIX B
COMMON ERRORS
Over the years, I have collected a list of common mistakes:
1) Hyphenated words-If the first word is used as an
adjective, no hyphen is necessary, e.g., first generation.
If the first word is a noun, then you need to hyphenate,
e.g., range-limited. If the second word is a gerund, i.e.,
the present continuous tense of a verb, then it is not
necessary to hyphenate, e.g., cell splitting.
2) Normally, integers less than ten are spelled out. Thus
one will write 搒ix cells?instead of ? cells.?Integers
larger than ten and fractional numbers are written in
arabic digits, i.e., 12, 5.6, etc. Fractional numbers are
considered plurals. Thus, we will say 搊ne meter,?but
?.5 meters.
?) In technical papers, there are usually symbols, and the
question arises as to which article to use in front of
symbols. Should we say a M/M/1 queue or an M/M/1
queue? The rule is the same as in regular writing without
symbols, i.e., if the word starts with a vowel, namely,
the letters a, e, i, o, u, you will use the article 揳n?
otherwise, you will use 揳.?However, we need to
determine how the symbol is pronounced. In the case of
M/M/1, we pronounce it 揺m-em-one,?i.e., it starts with
a vowel. Therefore, 揳n M/M/1?is correct. Compare this
with a B-ISDN network. In this case, the B in B-ISDN
is pronounced like 揵ee,?i.e., not a vowel.
4) The first time a symbol is used, explain what it means,
usually with the symbol in brackets, e.g., one will
write 揑ntegrated Services Digital Network (ISDN).
擲ubsequently, use the symbol only. This is in keeping
with the concept of conciseness.
5) Try avoiding negative words like 搉ot,?搖n,?搉on,
攅tc., as well as double negatives such as 搉ot invalid,
敁not uninteresting?as much as possible. For example,
use 搃nvalid?instead of 搉ot valid,?use 搗iolating
攊nstead of 搉ot satisfying.
?) The phrase 揳 lot of?is used for uncountable objects,
such as a lot of money. Do not use it for countable
objects, use the word 搈any?instead, i.e., say 搈any
users?rather than 揳 lot of users.?The same goes for 揳
large amount.?It is also used for uncountable objects.
7) Say 揼reatly improves?rather than 揾ighly improves
攐r 搇argely improves.
?) Say 揷ontrary to?rather than 搃n contrary to.?揅ontrary
to?is the same as 搃n contrast to?or 揳s opposed to.
?) The words 搘ork?and 搑esearch?are already in plural
form. Thus we do not say 揈xisting works in this area
?or 揚rior researches
?0) Do not use abbreviated forms like 揹on抰.?They should
be spelled out.
11) Avoid using multiple superlatives. Use 揵est?rather
than 搗ery best,?搊ptimal?rather than 搈ost optimal.
?2) The words 揻igure,?搕able,?搕heorem,?搇emma,?etc.
may be used as proper or common nouns. Proper
nouns must be capitalized. They are proper nouns when
a number or some other attribute follows them. For
example, we say, 揊ig. 1 illustrates ?and 揑n this
figure, we illustrate
?3) Do not start a sentence with "also.?Use words such as
"Besides,?"Moreover,?"In addition?instead.
14) Say "comprises?or "consists of?rather than "comprises
of.
擫I: HINTS ON WRITING TECHNICAL PAPERS 137
15) Semi-colons can be used to break up groups of objects.
For example, "Set A comprises numbers 1, 2, 3; Set B
comprises 4, 5, 6; Set C comprises 7, 8.
?6) Avoid repeated usage. Say " the storage required in
the first case is greater than that in the second case,
攔ather than " the storage required in the first case is
greater than the storage required in the second case.
?7) English and American spelling is sometimes different,
i.e., "colour?versus "color.?Try to be consistent
throughout the text.
ACKNOWLEDGMENT
The author would like to thank the people who have read
earlier versions of this paper and gave useful suggestions.
In particular, Dr. A. Choudhury of Lucent Technologies Bell
Labs, Dr. I. Khan of Qualcomm, Inc., and Dr. S. Sengodan of
Nokia Research Center gave comments in detail.
Victor O. K. Li (S?0朚?1朣M?6朏?2) received the S.B., S.M., and Sc.D.
degrees in electrical engineering and computer science from the Massachusetts
Institute of Technology, Cambridge, in 1977, 1979, and 1981, respectively.
He joined the University of Southern California (USC), Los Angeles,
in 1981 as an Assistant Professor, and became Professor of Electrical
Engineering and Director of the USC Communication Sciences Institute.
Since Sept. 1, 1997, he has been with the University of Hong Kong as Chair
Professor of Information Engineering in the Department of Electrical and
Electronic
Engineering, and Managing Director of Versitech Ltd., the University
Company. He has published more than 200 technical papers and has lectured
and consulted extensively around the world. His research interests include
high-speed communication networks, personal communication networks, and
distributed multimedia systems.
--
---- 归零 ----
※ 来源:.哈工大紫丁香 bbs.hit.edu.cn [FROM: 218.9.120.72]
Powered by KBS BBS 2.0 (http://dev.kcn.cn)
页面执行时间:203.923毫秒