Professional blog

Reviewing research communications
Review research communications to help you understand the state-of-the-art in your field.
Examine the common forms of research communication, and how can you undertake a
literature review.
Communicating research
Research communication advances the field of research.
It is important to communicate research outcomes in a way that is
understandable to peers and useful for a general audience. If you have
developed new techniques, methods, architectures, frameworks,
systems or algorithms, or discovered new insights, it’s important to let
people know.
As ideas are shared, others can build on them to advance a field. Each
piece of research can be seen as a ‘brick’ and as ‘bricks’ are laid, one
upon another, the field (ie, the ‘wall’) is built higher and higher.
Contemporary researchers in the field of Artificial Intelligence, for
example, enjoy decades of research resulting in literally thousands of
research publications. Companies can use the published research and
build their solutions on it.
Forms of communication
Research communications take many forms, including technical or
non-technical publications (or ‘papers’), presentations, datasets and
deployed systems. You may also share your research using patents and
demonstrators or proof-of-concept prototypes which are described later
in the week. You should choose to communicate in the way that best
suits your purpose for sharing.
The most common way for academics to share their research is
via research papers or publications in conference proceedings, journals,
book chapters and books or more technical books called monographs.
Much academic and industry research is published in order to share
new findings with people who may benefit from them.
Journals and conferences
Often, research papers are judged or reviewed by experts before they
are published. Some don’t make the cut.
There are conferences and journals that enjoy wide readership. They
are prestigious because they generally publish high quality research
which is well cited. However, they’re not the only places you can find
excellent research.
Conferences bring researchers in a certain field together to share the
ideas captured in their research papers. Typically, hundreds of
researchers submit their research papers to a conference, which has a
committee to select a small percentage of them to be presented. The
selected papers are then published in a conference proceedings.
Some examples of journals and conference proceedings on IT topics can
be found from a number of popular publishers’ websites, eg,
the Association of Computing Machinery (ACM) hosts and publishes
many journals, as do the IEEE Computer Society website, and Springer.
Publications for the general public
There is a wide range of research publications, ranging from more
technical publications for scientists, engineers and researchers, to less
technical and more accessible publications or newsletters for
communicating research to the general public.
Some examples of publications for a general audience that provide
useful IT related information include the MIT Technology
Review, Wired Magazine, Slashdot, and the Conversation.
Deploy or die: deployed systems
You can also share you research in the form of deployed systems.
It’s often said that academics must ‘publish or perish’. The MIT Media
Lab’s founding director, Nicholas Negroponte, famously set ‘demo or
die’ as an imperative for researchers to show proof-of-concept of ideas.
A more recent MIT Media Lab director, Joi Ito, went further to set
‘deploy or die’ as an aim for researchers in his TED talk.
Sharing data
Generating data can be a research outcome in and of itself. Data is
increasingly being shared because the ability to repeat, and to some
degree verify, published research is especially important in recent
times.
Some researchers are encouraged to publish the datasets associated with
their paper in a publicly accessible repository. There are many
repositories where researchers can upload datasets and make them
available to others. An example is Dataverse hosted by Harvard
University.
Data collected and used for testing are also often uploaded and made
available publicly. There are many repositories serving different areas
in IT, including CRAWDAD and Crowdsignals. These resources help
because data collection is usually a difficult and time consuming
process.
What is a literature review?
A literature review lets us know what research has come before us, and
what is still to come.
It’s a written account of a survey of ideas and developments in a field,
often based on reviewing, examining and critically reflecting on
published papers in the field. Often, a literature review is done at the
start of a research project in order to:
• understand the state-of-the-art in a particular area
• to get an understanding of the important ideas and background
knowledge of an area
• to identify gaps or opportunities for further research
• to compare work critically, eg, to argue why a certain piece of
research is original or better compared to other work.
When should you do one?
You would usually undertake a literature review at the beginning of a
research project to establish what has, and what has not, been done (ie,
the gap). Undertaking a literature review, whether for a university or in
industry, will help you size up and learn about current work in the area
and find solutions to a problem.
This will put you in a position to formulate and propose new research.
You might also undertake a smaller literature review in your paper or
conference presentation to demonstrate your research is original, and
compare it with other work.
Literature reviews can also be published in their own right.
What literature reviews provide
Often a literature review:
• provides a roadmap of an area
• presents a new perspective on a given area
• provides a classification (or taxonomy) of the published research
in a given area
• provides a chronological overview of how an area has developed,
eg, the timeline of the historical development of cars from 1900 to
today.
A roadmap, taxonomy, perspective, or timeline provide a conceptual
structure or a ‘lens’ to view the research in a given area. You could use
these to try to establish the place of a piece of research in a conceptual
structure.
Hence, a literature review should help us understand and form a
conceptual map of an area, and get up to speed with recent
developments in a field quickly. It is not just paragraphs summarizing a
set of papers, but attempts to link, connect, compare and contrast
publications in a field.
Good examples of literature reviews (sometimes called literature
surveys) in IT can be found in ACM Computing Surveys.
We will be looking at many examples of literature reviews later.
Reading research critically
Critically read research to refine your own ideas. Read papers, online material and patents
strategically and synthesise the relevant concepts.
How to read a research paper: structure
Understanding the structure of a research paper will help you to read it
strategically.
Research papers document the outcomes of research, together with the
data and reasoning involved in coming to its conclusions. Research
papers usually describe their research contribution, such as a new
concept, architecture, algorithm, experimental results or method. A
research paper is, then, different from a literature review paper.
There are different kinds of research papers. Some are shorter and just
contains a nugget of an idea. Others argue a position on a topic (often
called a position paper). Still others tend to be more complete and
presents research in a well-justified manner with the structure as
follows.
The structure of a paper
Most research papers have the following structure:
• title
• author(s)
• an abstract: a brief summary of the paper
• an introduction: the introduction often contains
o a broad overview of the field
o the aim of the research
o the main contributions of the paper, and
o a brief outline of the paper.
• related works: in effect, a small literature review. This section is
either after the introduction or before the conclusion
• the contents of the research: eg, describing the concepts,
methodology and discussion of results
• a conclusion, at the end of the paper
• a list of the articles cited by this research publication, sometimes
called the reference list or bibliography. Look at these carefully
because you might find more relevant articles
• appendices or links to related data may also be included.
How to read a paper based on structure
Understanding this structure can help you to read a paper strategically
and at an appropriate level of detail.
• read only the title and abstract to get an idea of the paper. Stop
here if it’s not relevant. If it is:
• continue reading the introduction and conclusion of the paper. If
the full details are relevant:
• read the full paper.
How to read a research paper: content
Some of the questions you should ask as you read are:
• What is the aim of the paper?
• What are the main contributions of the paper? What new ideas
are presented?
• How is the research contribution or claims justified?
o What sorts of experiments were conducted and how were
the experiments set up?
o What data was obtained and how does it justify the claims
made?
o Were the experiments adequate? Did they gather enough
data to support the claims?
• Was data interpreted correctly?
• Was the reasoning in the paper sound?
• What were the assumptions made in the research? Are the
assumptions valid or realistic?
• How new or significant is the research idea or contribution?
What are the implications to the field of the research presented?
• What are the novel aspects of the work?
• What are the shortcomings of the work? What are strengths and
weakness?
• How can the work be extended or shortcomings dealt with? Does
the paper encourage future work in the area?
• Are the results generalisable or applicable to another context? Are
there improvements that could be made on the work?
The questions above encourage you to take a critical view of the work
being presented. A critical view looks for strengths and virtues, as well
as potential weaknesses.
Often, it is important to make notes as you read the paper and create a
short summary (at least a few sentences), perhaps in answering the
some of the questions we’ve just taken a look at.
It’s not uncommon to read through the paper several times.
We will look at many examples of research articles later.
On reading online
You can find some great material for the general public published
online.
No matter where you source your material from—whether it’s a
research paper or a website—you need to read critically.
Sometimes research is shared online in newsletters, magazines such
as MIT Technology Review or articles like those is Wikipedia.
Wikipedia is an important resource with a great deal of useful
information.
Wikipedia articles often have links and references to other important
resources. You can use these resources to find out more on a topic, or
judge the content of the community authored articles.
Most papers published in journals and conference proceedings have
been reviewed by several technical experts before they are allowed to
be published. In general, they are more valid and verifiable than
websites.
Apart from judging the contents, it’s also important to check:
• the author of the article
• the institutional authority behind the article (which organisation
is hosting the site and what is their interest in the information?)
• date of the web article
• the references to other articles the web article provides.
Patents
Patents help us find out more about commercial IT R&D.
If someone’s research has commercial value, they might choose to
protect their intellectual property by filing a patent on it, rather than
publishing. Looking at patents can be a good way to see state-of-the-art
research designs, and also to see some of the strategic technological
decisions that a company is making.
Many large companies have numerous patents. Take a look at examples
of IT patents from:
• IBM
• Amazon
• Google
• Apple
What is a patent?
According to IP Australia:
A patent is a right that is granted for any device, substance, method or
process that is new, inventive and useful.
A patent is a legally enforceable right to commercially exploit the
invention for the life of the patent.
Moreover, a patent will:
• give you the right to stop others from manufacturing, using
and/or selling your invention in Australia without your
permission
• let you license someone else to manufacture your invention on
agreed terms or take legal action against people who are using
your invention without your permission.
How long are patents enforced for?
• Pharmaceutical patent: up to 25 years.
• Standard patent: up to 20 years.
• Innovation patent: 8 years.
You may also file a provisional application before applying for any
patent to make sure you have a record of when you had an original
idea.
Who can own a patent?
Patents are owned by the inventor, a person who has legally purchased
(or been given) the invention or the company, organisation or other
employer of the person who made the invention in the course of their
normal duties.
Many companies rely on their research being patented and then
licensed to other companies as a source of revenue. An interesting
example of a lucrative patent is the patent on the Australian invented
WiFi that billions of people use every day.
International patents
An Australia patent provides protection in Australia. You might need to
file a patent in each country that you want protection in, or file an
international application under the Patent Cooperation Treaty for
protection in 151 countries (administered by the World Intellectual
Property Organization (WIPO)).
Syntopic reading
It is important to read more than one perspective on a topic to obtain a
balanced view.
A literature review typically surveys ideas from different research
papers, and different viewpoints, providing a comprehensive overview
on a particular area.
The idea of reading multiple perspectives on a given topic is sometimes
called syntopic reading (Adler & Van Doren 1972), roughly meaning a
synthesised understanding of a topic generated by reading many books
and papers.
The topic that you have in mind becomes the primary focus of your
reading. For example, you might want to learn about the topic ‘Internet
of Things’. You might have questions, such as:
• What is the Internet of Things?
• Why the Internet of Things?
• What are applications in the Internet of Things?
• What are the issues and challenges for the Internet of Things?
(In fact, we will look into Internet of Things at greater depth in a later
course.)
To get answers to these questions, you might then read, not just one,
but several books and articles.
Each author provides their perspective on the topic. It’s up to you to
identify if there are ideas in common or contrasting views. The
important thing is to analyse and connect material from different books
and papers.
For example, suppose we want to know about the topic ‘SelfAssembling Robots’. You might form a number of questions such as
‘what is it’, ‘how does it happen’, ‘what are the algorithms involved’ or
‘what are the engineering design issues’, etc. You could read books and
papers on each topic with these questions in mind.
Writing a research proposal
Apply the Design Science Research Methodology to research planning, by writing a research
proposal.
Writing a research proposal
Soon you’ll be asked to write a research proposal. Before you write it,
you’ll need to know what’s in one.
A research proposal is a document that demonstrates you have
systematically thought through the research you are proposing to do.
They are often done early on during a research degree such as Masters
by Research or doctoral studies, when applying for research funding or
proposing a project in industry.
An expert from the institution supporting your research will use this
proposal to evaluate whether or not to support your research.
The structure of a research proposal
While the structure of a research proposal can vary, typically it
includes:
• Project Title.
• Background, Related Work and Justification: this section sets the
scene by providing the background and context for the research.
This is in fact a literature review of the area of the project.
• Aims and Objectives: this states the aim of the research project—
what the projects aims to discover, prove, demonstrate, evaluate,
analyse, examine, investigate, or develop. Sometimes, a research
hypothesis is stated. Expected outcomes and why the research is
important can be emphasised here.
• Methodology: this section describes how the aims of the research
will be achieved, including:
o what conceptual framework or theory will be applied
o what experiments will be set up and how, including any
applicable details of hardware or software to be created
o what kinds of data will be acquired and how will the data
relate to achievement of the aims
o if there will be a need for user testing and if so, how will it
be conducted
o what reasoning will justify the claims you make
o how the outcomes will be evaluated
o possible barriers, if any, to success and how they might be
overcome.
• Timetable: a chronologically placed plan of action.
• References: research papers mentioned or cited in the literature
review.
Here is an example of a research proposal by Daryush Mehta from MIT
Media Lab. Please note how the document has the above components.
Design Science Research Methodology (DSRM)
There are a lot of similarities between the Design Science Research
Methodology (DSRM) described in Week 1. This is because a research
proposal should be guided by a methodology that typically follows the
DSRM.
It’s also worth noting that a research paper is usually also compatible
with the DSRM. A paper, of course, includes information not just on
intentions, but outcomes.