Transcript Slide 1

Writing (HCI) Research Papers
First developed by Duke Hutchings November 28, 2005.
I was greatly aided by
- Gregory Abowd
- Keith Edwards
- Beki Grinter
- Elaine Huang
- Jeff Pierce
Focus: conference papers
(though most topics apply to journal paper writing too)
Typical Paper Layout
Abstract – summarize your results and contributions
Introduction – motivate the reader and frame your goal or problem
Related Work – indicate what others did prior to your work
Your Work – state what your work is
Future Work – what should be tackled next?
Conclusion – summary of work and implications
References – listing of past and related work
Paper-writing is a “people business”
Conference structure:
paper chair(s)
paper committee
reviewing system
You need to make sure that…
- the paper gets into the right hands
- you satisfy your reviewers
- you are aligned with the right conference
Abstract and the Paper Chair(s)
You need to make sure that…
- the paper gets into the right hands
In the abstract, briefly state
- the general research area that you address
- the main result(s) that you produced
- the method(s) used to achieve the results
- the impact your result(s) will have on people
Don’t put yourself “halfway out the door”
An example abstract (is this good?)
(see http://www.cc.gatech.edu/~hutch/papers/hutchings2004revisiting.shtml)
Most modern computer systems allow the user to control the space
allocated to interfaces through a window system. While much of the
understanding of how people interact with windows may be regarded as
well-known, there are very few reports of documented window
management practices. Recent work on larger display spaces indicates
that multiple monitor use is becoming more commonplace, and that
users are experiencing a variety of usability issues with their window
systems. The lack of understanding of how people generally interact
with windows implies that future design and evaluation of window
managers may not address emerging user needs and display systems.
Thus we present a study of people using a variety of window managers
and display configurations to illustrate manager- and displayindependent space management issues. We illustrate several issues
with space management, and each issue includes discussion of the
implications of both evaluations and design directions for future window
managers. We also present a classification of users’ space
management styles and relationships to window system types.
Reviewers
You need to make sure that…
- you satisfy your reviewers
This can take an unlimited number of forms…
- did you talk about your reviewer’s work?
- does your reviewer, an expert, understand your approach?
- does your reviewer, a “novice,” understand your approach?
- does your reviewer like the topic, style, method, etc.?
- do your terms and descriptions match the reviewer’s?
Let’s focus on writing about related work
Related Work and Reviewers
Why do people often fail to mention related work?
- it sounds like the work has already been done
- it looks like lots of people are already working on it
- it makes the work sound more unique
Actually, a thorough related work section accomplishes those
objectives. A weak related work section allows your reviewer to
immediately dismiss you as uneducated and ignorant.
When mentioning related work, don’t bash it
- your reviewer may have done that work
- your reviewer might think you’re a jerk
A better approach is to show how your work differs from past work
and avoid any characterization of that past work as weak or bad.
An example opening to related work
(see http://www.cc.gatech.edu/~hutch/papers/hutchings2004display.shtml)
More users are opting for multiple monitor systems, and initial lab
research indicates that multiple monitor systems can help users
be more productive [4] but that multiple monitor systems could
stand to gain from advances in hardware and software design
[17]. These important findings motivate field work such as ours in
order to understand actual management practices that people
employ..
Related Work – don’t overdo it
You have only so much space in which to describe past work.
Pick 12 – 20 pieces of the most relevant related work and
compare or contrast pieces of that work with your own.
Generally,
- 8- is too little
- 25+ is too much
Note: for journals, you can spend more time on related work
A quick note on “et al.”
First “et al.” is short for et alia, Latin for “and others.”
Second, whenever possible, avoid the use of this phrase. Would
you like to be referred to as “some other person” when an author
references your work? In general, “et al.” should refer to papers
with 4 or more authors.
Third, never under any circumstances use “et al.” in the
references section of your paper. List every author from the
paper, whether there are 2 or 20.
Fourth, it’s not et. al. (two periods). It’s et al. (one period)
Related Work – where is it?
Um… I new at this. How do I find out what the related work is?
- Ask someone (actually, ask several people)
- Use the ACM Digital Library or IEEExplore
- Use Google Scholar or CiteSeer
- Check DFAB 
Reviewing – like your 6750 project
Most conferences use the same reviewing format from year to
year or even publish the reviewing for to be used for the current
year. Find out what the typical reviewing form looks like and write
your paper accordingly.
Picking your spots
You need to make sure that…
- you are aligned with the right conference
Positive answers to these questions bode well for you:
- Does this conference have a history of papers on my topic?
- Does this conference have a history of papers using my
approach or method?
- Does this conference have a recent history of my topic?
- Does this year’s conference committee have an expert on
my topic?
- Is this conference becoming interested in my topic?
A note about conference themes… typically only a small number
of papers actually address it, so don’t target it.
General HCI Conferences
You need to make sure that…
- you are aligned with the right conference
CHI
Interact
HCII
GI
OZCHI
HCI
LAHCI
“The” conference
Another International HCI Conference
Another International HCI Conference
A Canadian HCI Conference
An Australian HCI Conference
A Bristish HCI Conference
A Latin-American Conference on HCI
Specialized HCI Conferences
You need to make sure that…
- you are aligned with the right conference
UIST
CSCW
Ubicomp
InfoVis
AVI
DIS
DUX
CUU
Pervasive
IUI
Mobile HCI
interface software and technology
collaborative and cooperative interfaces
ubiquitous computing
information visualization
visual interfaces
design
design
universal usability
pervasive computing
intelligent interfaces
mobile HCI
Conferences with some HCI
You need to make sure that…
- you are aligned with the right conference
C&C
Assets
CVE
DARE
DPPI
ETRA
MM
NPVIM
PDC
SV
cognition and creativity
assistive technology
collaborative virtual environments
augmented reality
design
eye-tracking
multi-media
information visualization
participatory design
software visualization
Picking your spots
You need to make sure that…
- you are aligned with the right conference
Papers get accepted for a variety of reasons
- they introduce a brand new topic
- they make significant improvement to an old topic
- they confirm an older result using a new method
- they open a very wide line of new research opportunities
- they are controversial or provocative (but factual)
By becoming familiar with a set of conferences, you can
determine how your work best fits with a specific conference and
how it is situated within the community of that conference
Various paper-writing tips
Two good sites to have in your bookmarks:
http://swig.stanford.edu/~fox/paper_writing.html
http://www.alice.org/Randy/raibert.htm
Where do I start?
Beginning authors can have a hard time understand how to get
started. One of the best approaches is to follow a successful
model, as provided by another author’s past work.
- Find an oft-cited paper that is similar to your work
- Follow the leader: intro, related work, your work, etc.
Beware: some aspects of that paper may have been negatively
reviewed or received by the community-at-large
Try to find a local expert who can not only provide the paper
but also provide the reviews of the paper (including any
versions that were not accepted at a conference)
Trap: not enough detail
The authors do not explain how their 20 participants were chosen.
Trap: unjustified claims
The authors mention that these participants "are representative of
information workers in any company or group", which they provide
no evidence for.
Trap: careless use of language
At one point in the paper the authors state that "We witnessed five
different ways that participants switched among windows." Yet,
they never mentioned that they observed users interacting with
their system.
- The difference between “does,” “will,” and “can” is dramatic
Problem Statement – don’t overdo it
One of the contributors, who has reviewed many papers, had this
to say about authors’ problem statements:
Too many papers try to make it sound as though they are solving
a problem for which the current solution is completely broken
For example, if you're trying to make a cellphone that will sense if
you're in a meeting and shut its ringer off automatically, go ahead
and say that remembering to turn it off is a problem and that it can
be annoying and disruptive when a phone rings during a meeting.
Don't say that this is a breakdown that makes cellphones
practically unusuable and is leading to the decay of society.
Using images and screenshots
Images and screenshots can go a long way in helping to explain
visual concepts or dense amounts of data.
It may be hard to tell where images are best used though because
as an author, you are intimately acquainted with your interface or
data. Get a friend to read through and ask where pictures would
help explain how an interface works or how results compare.
Get proofreaders
Research paper proofreading takes two forms:
- traditional proofreading: grammar, spelling, clarity, etc.
- content proofreading: will this paper be accepted?
While any colleague will do for traditional proofreading, more
senior students and professors will have a better grasp of what it
takes for papers to be accepted. As a result:
Start early:
- start writing early
- start locating proofreaders early; they’re busy people
Misc. traditional proofreading tip: read backwards.
Going backwards: a possible approach
One colleague has said this about paper-writing:
I have started to shape my research such that I “write” the paper
before I do the research. Since the end result of research is
typically a conference paper, I work backwards from that. I figure
out what I want to be able to say (usually the answer to a research
problem I expect to arrive to), then figure out how I can structure
my research to arrive at that answer, then do the research. It
makes writing the actual eventual paper very easy.”
It takes awhile to grasp
Don’t feel bad if your first attempt goes horribly wrong. It takes
awhile (mistakes) to understand the pitfalls and gold mines.
Don’t necessarily feel good if your first attempt goes well. Almost
always reviewers state concerns about the very best work. Take
the concerns to heart and try not to make those mistakes again in
future writing.
So…
Just Do It! The earlier you start writing, the earlier you can move
past beginner’s mistakes.
I’m running out of space!
Frequently you find that you cannot fit what you want to write in
the allotted amount of space. This could be an indication that you
are not being succinct enough. If not, don’t forget these tricks:
- automatic hyphenation (Word: tools>language>hyphenation)
- spacing between lines and sections
- typically no one cares if the references are in 9pt instead of 10pt
Don’t forget though… with so many submissions coming into a
conference and only a limited number of reviewers available, it’s
easy to automatically reject papers that are improperly formatted.