Information Technology Reference
In-Depth Information
conferences where relevant work appears, and papers with lists of references
to explore.
Follow up the references in promising research papers. These indicate relevant
individuals, conferences, and journals.
Browse the recent issues of the journals and conferences in the area; search other
journals and conferences that might carry relevant papers.
Search the publisher-specific digital libraries. These include publishers such as
Wiley and Springer, and professional societies such as the ACM and IEEE. There
are also a wide range of online archives, in particular
www.arXiv.org
.
Most conferences have websites that list the program, that is, the papers to appear
in the conference that year. Within a conference, papers are often grouped by
topic—another hint of relevance.
Consider using the citation indexes. The traditional printed citation indexes have
migrated to the Web, but in practice their value for computer science is limited, as
only a fraction of publications are included.
Go to the library. The simple strategy of having similar material shelved together
often leads to unexpected discoveries, without the distractions that arise when
browsing the Web.
Discuss your work with as many people as possible. Some of them may well
know of relevant work you haven't encountered. Similar problems often arise in
disparate research areas, but the difficulties of keeping up with other fields—the
phenomenon sometimes characterized as “working in silos”—mean that people
investigating similar problems can be unaware of one another.
The process of search and discovery of useful papers can be thought of as a form of
learning. Each paper or page that you find should refine your understanding of the
terminology, help indicate which papers are significant, suggest new directions to
follow up, and further clarify your criteria for whether a paper is “in” or “out”.
Occasionally there are several versions of the same paper: a preprint in an online
archive, a conference version, and a journal version.
2
You should use the version that
the author appears to regard as definitive; this will usually be the polished work that
has appeared in a journal.
Take a broad definition of “relevant” when searching for papers. It doesn't just
mean those papers that have, say, proposals for competing methods. Does the paper
have interesting insights into other research literature?Does it establish a benchmark?
Have the authors found a clever way of proving a theorem that you can apply in your
own work? Does the paper justify a decision to not pursue some particular line of
investigation? Other people's research can have many different kinds of effect on
your work.
Finding
all
relevant work is hard; for example, exhaustive professional searches
across the medical research literature can take months of full-time work. But finding
2
As I have noted elsewhere in this topic, the practice of revising a conference paper to create a journal
submission was once common, but, for refereed conference papers that are available online from
a major publisher, is now infrequent—and is increasingly discouraged. However, where multiple
versions exist, they need to be cited correctly.
