Informatics Research Methodologies
Guide to Writing Grant Proposals
Most informatics projects entail writing a computer system
for some task, or applying an existing computer system to a
task. We will assume that context below, but some of the points
will also be applicable to other kinds of projects, eg purely
theoretical ones.
- What task will your system be applied to? Why is this task
(a) important and (b) challenging? If some technical background
is needed for a non-expert audience to understand this task then
give this background as succinctly as possible. You can assume
your referees are expert informaticists, but are not specialists
in your particular research area.
- What techniques will be used in your system building? These
might include techniques for software engineering, requirements
capture/knowledge acquisition, data/knowledge representation,
architectures, algorithms, theoretical analysis, experimental
methodology, etc. Will the techniques be suitable as they stand
or will you need to adapt, extend or generalise any of them?
Explain any such planned modifications. Give any technical
background needed to understand these techniques as succinctly as
possible.
- What claim(s) will you make about your system? Will this be
the first system to tackle this particualr task? If not, how do
you expect your system to compare to its rivals? You could
compare it on dimensions of behaviour, coverage, efficiency,
usability, dependability, maintainability, psychological
validity, etc.
- What evidence will you produce to support your claim(s)? Will
it be theoretical or experimental? Anticipate the mathematical
techniques you will use in any proofs. Give the design of any
experiments.
- Who will benefit from this project? Beneficiaries might
include fellow reseachers working on similar problems or using
similar techniques, potential users, product suppliers who might
incorporate your ideas in their products, etc.
- Identify the different "workpackages" required to specify,
implement, evaluate and disseminate your project. Break the
implementation into modules. For each workpackage give: its
duration, its dependencies on other workpackages, its
"deliverables". Deliverables might include a specification, a
module of the complete system, experimental or theoretical
results, a publication, manual, etc. Draw a "diagrammatic project
plan" relating these workpackages.
- How will you disseminate the results of your research
project? Some of the options are: publication in internal
technical reports, workshops, conferences, journals and books;
making publications and software available over the web; talks at
departmental seminars (internal and external) and conferences;
tutorials at conferences; etc.
- How will you manage the project? Regular, informal
supervision meetings are sufficient for small one researcher
projects, but more formal management is needed for multi-person,
multi-site projects, especially ones with a mixture of academic
and industrial partners.
- What resources will you need? The number of person months
should equal the total duration of all the workpackages. How many
people and what kinds of skill mix do you need? They don't all
have to work for the same period. What equipment will you need:
just workstations/laptops, or do you have special needs: chip
fabrication, robots, speckled computers, high-performance compute
servers? Where will you need to travel: conferences, lab
visits?
Alan Bundy
