The goal in trying to improve the web site is to show off the work of the department. This document describes how to write a short piece (extended abstract length) that will allow you to do that. We don’t expect what you produce to perfectly fit the guidelines described here, just for you to keep this specification (and the philosophy behind it) in mind as you compose. If you have any questions at all about this, please don’t hesitate to contact Sunny Bains <s.bains@imperial.ac.uk>.
Audience
You should not think of your research project summary as a technical paper, but as an advertisement for your work and your expertise (so that people want to read your publications, collaborate with you, and give you money). The main thing is to get across the importance of what you’re doing, how you are doing it, and the expertise that you and your group have that make you qualified to be able to succeed at it.
However, we don’t want it to be too fluffy. When people within your own field read about the project, we want them to have some idea of how your work differs from your competitors, which may involve getting a little bit technical and/or referring to your publications.
Essentially, you are having to address three different audiences at once. First is a general technical audience (we’ll call it G) consisting of people from very different disciplines: if you are an computer scientist, it may be useful to think of a reader who is a microbiologist. This class of people is important because it includes industrial managers, sponsors, civil servants etc.: people who have technical backgrounds but don’t necessarily have any knowledge that is relevant to your particular work. These are often people who hold purse strings and make employment decisions.
Second is the subset of readers within your general discipline (we’ll call this D): other computer scientists. They have degrees in CS (possibly from many years ago), but their work may be drastically different to yours (for instance, consider someone in machine vision trying to talk to someone working in quantum computing). Finally, there are those people who are actually specialists in your field who want to know the meat of your technological contribution (we’ll call them S).
The formula to please all your audience is very simple. A typical 500-word project description (plus one picture with a short caption and as many references as you want) will normally consist of about five roughly 100-word paragraphs that should be addressed to the following audiences: G, G, D, S, G. For fuller details, the actual content, see the outline—with examples—below.
Finally, as you go along, please think carefully about the publications you want to reference and also, if you can, about any pictures that could be used to illustrate your work. If it’s a diagram and you’re concerned it’s ugly, remember we can have it redrawn for you: the idea is to choose something that really communicates what you’re doing.
Outline
Name of the project with important acronyms spelled out (General technical audience)
Keep it short (ten words or fewer), straightforward, and as free from jargon as you can.
FUSE: Floodplain underground sensors
Short description (General technical audience)
This is not an abstract, just a sentence or two summing up what the project is about: the whole thing should be no longer than 35 words. It is going to show up in lists and will always be seen with the name of the project above it. For example:
We are creating enhanced algorithms that perform pattern matching for applications like medical screening and biometric identification. Novel coding allows them to operate orders of magnitude faster than current approaches and with a lower false-positive rate.
Paragraph 1: The problem (General technical audience)
Give the context of your work for a non-specialist. To what general field(s) does your work apply, and why is this field important? What could be achieved in these fields and what issues are holding back this progress? Remember to spell out all acronyms the first time you use them, and to explain all jargon terms that aren’t well understood outside your field.
Please write the main text in the active voice where possible, using the first person or first person plural (“we have developed…”). For instance:
Having to use ionizing radiation for body scans has many disadvantages, particularly related to the safety of both medical staff and patients. It would be ideal if we could use visible light instead: it is simple to generate, poses little risk to humans, and can captured easily using electronic cameras. Unfortunately, however, diffusion of light rays through body tissue mean that the images that emerge are unclear…
Paragraph 2: The set up (General Technical Audience)
From the issues you described in the first paragraph, now pick out the ones that directly relate to your work. How have people tried to solve this/these in the past? Why have these solutions fallen short? What is (briefly) your new solution? For example:
There have been many approaches to the development of micromechanical structures with very high aspect ratios. Most of those in two dimensions have problems of low-yield. For instance, with three-dimensional techniques, such as {techniques and problems here}. To get around these issues, we have been working on a new approach called hybrid structure formation (HSF), that attempts to incorporate the advantages of both and the disadvantages of neither.
Paragraph 3: Your approach (Audience in your discipline)
Having mentioned your approach in the last paragraph, you should now explain the basic concepts behind it and how it works. Here you can be a little more technical, but if you use words that can’t be looked up in a basic scientific dictionary, add some explanation. For example:
Neuromorphic engineering—the building of brain-like structures in silicon—was originally conceived by Carver Mead at the California Institute of Technology.6 It is based on the idea that it is more efficient to use the physics of electronic devices to implement functions directly, rather than to simulate these functions using digital algorithms. This is a particularly advantageous approach for building neural systems as the functions that transistors perform naturally are qualitatively very similar to those in biological neurons…
Paragraphs 4: What you’ve done (Audience in your discipline for paragraph 4, moving to a specialist audience for 5 and 6)
This is the most straightforward section of the article, and the one that is least likely to be a problem. Just very briefly outline what you’ve done and link to the papers where people can find out in detail what this has involved. An easy way to think about this is simply to use one sentence to describe the work that’s been published in a single paper, and to string all the papers/sentences together into a narrative. For example:
Our work has shown that, to maximize performance, AI in embodied systems must be optimized for the specific sensors and actuators of each individual body: not just as designed, but as they work in reality.1 Because sensors and actuators are not digital but analogue, this means that all machines, even built to the same specification, will be different.2 This means that the final optimization must take place through learning on the actual robot or other machine that the intelligence is to control.3 To achieve this we have developed an approach called…
Paragraph 7: Summary and further work (General audience)
Without restating from scratch and explain how the work you’ve just described has changed/progressed the problem you described at the top. Then tell us how you think you can make even further progress. For example:
To make the types of three-dimensional structures we need for next-generation batteries and other devices, the use of surfactants as templates seems promising. So far our results have shown that we can easily make structures at the right scale (5-10nm), in the right types of patterns (such as hexagonal cells), and that are solid enough to be viable in real devices. Our next step will be to show that the new materials can be fabricated in large-enough slabs to be useful in macroscopic products.