I am currently undertaking a masters in Advanced Computing Science at the University of Glasgow, which is heavily research based. Due to this I've been reading a lot of research papers and have developed my own style for reading them which I will share with you here.
First get a good
.PS copy of the paper in question. Print the document, I usually go for double sided to save space and 2-up if it's a particularly long paper.
If the paper has a lot of pictures or diagrams take the time and expense to print it in colour. You'd be amazed at how much more a colourful document will hold your attention over a grayscale one, also detail will not be obscured due to poor colour choice from the author.
If you're using any sort of bibliography database, and you really should be - BibTeX is my preference, then you'll be generating keys for all the documents you enter into it. Write the paper's key on the top of the paper along with the year it was published.
This saves time if you're citing the paper later, rather than having to search though the database, usually a flat file, to find the title or author of the paper, you simply have to glance at the first page.
Leave space at the top of the paper for any general comments you have about the paper overall. Don't write a critical review, just a quick note which you will be able to use to identify the contents of the paper quickly. Another useful thing to add to any paper is a list of keywords. Many conferences and journals add these to papers but your own keywords will have more meaning to you than the general ones produced by the publishers.
Once you have read the paper look through the list of references, highlight any that you think are potentially useful and try and find them, this is possibly the path your research will follow and maybe the next paper you should read. Once you have a copy and thus a key for the paper, due to your corresponding bibliography entry, note the key next to the reference. This allows you to quickly find out if you have the paper at a later date for example when rereading the paper for other research and allows you to quickly assess which paper it is.
Reading the paper
Something we've had knocked into us on the MSc course at Glasgow is how to read a research paper. The general consensus is a 5 stage process:
- Read the abstract - this should roughly tell you what the paper is about, what the problem addressed is, what the authors did and what they found.
- Read the introduction and conclusion - skipping to the end of a book usually gives away the plot, skipping to the end of a research paper will help you understand the ideas in the paper as it will contain a concise version of everything the author has said. You should have a fair idea of what's coming from reading the abstract anyway. Reading the start and end of the paper will give a better insight into what the paper is about and help you decide if it applies to your area of research or not.
- Read the whole paper - jumping past anything which you get stuck on. Read all the details quickly, glance at figures, tables and equations but don't over invest time in trying to understand them. Reading the whole paper quickly will give you an overall feel for the argument put forward in the paper.
- Re-read the paper - this time going into the fine details and trying to understand the parts of the paper which eluded you the first time round. If necessary look at some of the supporting material, consult books, websites, peers, colleagues and so forth. Make notes in the margins of the paper, most papers are typeset in either one or two columns, this gives space on either side of the text to annotate it - make notes, reference other works, point out problems e.t.c.. This is where it is useful to have a physical copy of the paper. Highlight key areas of the paper using whatever means you wish, I will go on to talk about this in the next section.
- Summarise - write your blurb on the front page of the paper and then try and write a page summary of the whole paper. Aim to include the main points of the paper, the arguments, methodologies uses, results and conclusions which you feel relate to your research or may do in the future. This is useful if you need to grasp the idea of the paper quickly at a later date, don't rely on remembering about the paper - you will forget the details.
Everyone has a different method for marking parts of the paper they are reading. I prefer to use highlighter pens as they do not obscure the detail or text in the document unlike circling or underlining.
Most papers, I find, tend to break down into four main sections:
- Introduction and related work
- Problem definition and intended methodology
- Experiment details and results
- Discussion and conclusions
Each of these sections I match to a different coloured highlighter pen, the same for every paper to make it easier to find sections of the document. I use:
- Yellow - Introduction
- Orange - Problem definition
- Green - Experiment details
- Red - Conclusion
Obviously some papers don't follow this general pattern, such as survey papers, in this case I use the yellow and red pens as before, for the introduction and conclusion. I then alternate between the orange and green highlighters for different sections or ideas within the paper body.
I urge you to keep all research papers after you've read them. Keep a digital copy, but also keep the printed copy which contains your annotations. It is also useful to keep a printed copy of your page summaries with the corresponding papers for a quick reference to the paper.
As the physical copy of the paper is likely to be the only annotated version of the paper you have make sure you don't loose it. As I've mentioned you will forget the details and annotations can be a low cost solution to jogging the memory.
Get a box file, put the papers and summaries into it, organise them in any manner you wish but just keep them.
All this seems a lot of work just for reading one research paper, but the this time is an investment which will be paid back in droves later.
Whenever you are required to cite papers or give related/supporting work sections in reports or your own papers the page summaries will provide you with the textual descriptions of the content that you will likely use. Re-writing a section on a particular paper five times, and consequently re-reading the paper five times, will take up more time than annotating and summarising the paper originally.
I enjoy reading research papers, as long as they're not too technically heavy, but reading research papers can take a long time. I've found that using a formulated and structured approach to reading, annotating and summarising the papers make the process easier and quicker, this is why I'm sharing my guide with you.
That's all I have to say, if anyone has anything to add, or their own unique style for reading papers, feel free to add a comment detailing it.