I am new to my field (cell biology / immunology) and I would like to ask about the best practice to annotate or summarize research articles and reviews for long-term retrieval. When I start reading into a new topic, I end up highlighting almost each paragraph of the paper/review. This doesn't make any sense, and it is actually making me slow and frustrated and it is holding me from reading more. Could you please share your thoughts.

Update: although two very important points of how to approach new papers and how to tag them have been addressed in the answers so far, my main concern was about the new acquired details which I can't remember a few months later. For example, I read a paper about JAK/STAT signaling two months ago, now I can't remember for example which JAK attaches to which cytokine receptor and interacts with which STAT. I need to skim through the paper again to find this piece of information. So, what I am looking for is an approach to accumulate the new knowledge and to facilitate reviewing them without the need to skim through the paper again and again. I know this will come with experience, but at the beginning there are too many new facts to keep track of.

  • Thank you for you feedback, I totally agree. I hope one of the moderator can transfer the question to academia.se. Please all excuse my ignorance of the different boards at stackexchange.com I am really impressed by the collaboration here.
    – Live
    Aug 12, 2012 at 20:44
  • @Live just double checking that Academia are happy to take it then we'll get it moved across for you straight away.
    – Rory M
    Aug 12, 2012 at 23:29

8 Answers 8


I find it important to be as active as possible in your reading, especially for papers close to the questions you are working on. If you are only highlighting, then this is largely a passive activity, and does not help much with retention or even later rereading. If you are making notes, it is a little bit better, but still largely passive.

For important papers, I try to read the paper and then write up a quick summary for myself of the key ideas relevant to my interests: a sort of mini-review or personalized-abstract of the paper. For papers that don't contain important tools I use, but contain facts I have to cite I usually include a section that lists in what settings I would expect to cite this paper. This helps a lot at the writing stage. These little summaries of the paper are usually too rough and critical to be useful to others, so I keep them private.

I usually organize these summaries in a little personal wiki like Tiddly Wiki where I can tag and do full text search as well as include links to the original PDF of my computer. Each article gets its own tiddle and is tagged according to topic. Other useful tools are available in:

Is there any efficient non-linear note-taking software?

For some heavy papers (say important math papers) the reading has to be even more active. Usually including reproving the theorems and sometimes presenting these proofs to others in my group.


As gkadam said, there is no absolute best way, and finding the way that suits you the best is also part of the learning process, and it can strongly depend on how your memory work. For some people, it's enough to just add some tags or keyword on the paper, allowing them to retrieve the information later, while other people needs to rewrite some parts in order to remember it.

For instance, the first time I read a paper, I do it without a pen, because I want to get a "feel" of the paper first, and if there is a point that I don't understand or that I think it's really important, I just leave it for later. I then read some of the references of the paper to see if I can't find a similar idea explained a bit differently (typically when there is an extended version of the paper I'm reading). After reading several papers, I start to have a good idea of the problem, and what part of which paper is the most important, and at this point I can read again, in detail, with a pen and some paper, to go through the tough parts of the paper.

The time you "lose" by reading the same paper several times, you gain it by focusing on the most important parts and by "skipping" the redundant ones.

  • Very interesting strategy to tackle a tough paper, and to know the context of the paper/problem being addressed. Many thanks indeed.
    – Live
    Aug 13, 2012 at 11:32

Since I read papers electronically, and save the PDFs to my drive, I put the most useful part of the paper, along with the first author's name, journal abbreviation, and year in the file name. It means I have long crazy file names but I always know why I saved that paper even if it was years ago. For example:

Chang_ChemRev_2009 - polymers for photovoltaics - PPV on p. 5874 and figure 3.pdf

  • This is a useful hint, I have not thought about before. It's indeed a "crazy file names" but I am depending on a Reference Manager "Sente" which may not allow to rename the file the way you describe it (understandably), however I usually add a comment explaining the reason the I added the paper for which serve the same purpose.
    – Live
    Aug 13, 2012 at 11:35
  • I save the file separately from my reference manager, and then link the reference to the PDF.
    – Ben Norris
    Aug 13, 2012 at 12:40
  • In related news, JabRef is a reference manager that allows you to rename the paper as you wish, as long as the name contains your "canonical abbreviation" for the paper somewhere. Aug 13, 2012 at 16:19

I'm using jabref as reference manager (I read the comment that you use Sente, which I don't know). Jabref allows me to assign a paper to different groups, and also to set up grouping in a hierarchical way.

In addition, I can put notes that are searchable (however, usually I don't do that but write them on the paper). I guess that's what I'd do with your JAK/STAT things. Sometimes I write the important stuff on a new sheet (if the margins of the paper are too small) and file that together with the paper.

I'm still going mainly with printed papers, as I can have a whole bunch of them besides each other and keep the overview which says what while writing.


I tried to copy-paste the key ("most relevant") sentences from the paper in a form of list, by using just text editor (to make it searchable). Drawbacks: 1. after 30-40 papers this text file started to be really big and difficult to search. 2. when I come back to this paper again the "important" parts are, in many cases, different from what was annotated from previous reading. 3. unable to sort the notes easily to categories. 4. in many cases even several sentences are not enough to express the finding.

Then I decided to excerpt the facts from the article pretty much the same way as they are prepared for Beilstein chemical database. I read an article briefly for the first time to roughly split it into logical pieces. Then I read it thoroughly sentence-by-sentence and make sort of reference card for every fact originally described in it (no information from references is included even though it may be relevant), although facts that go with "data not shown" count. In the reference card I have special fields to specify all available information on how particular fact was discovered including methods, experimental model etc. Every card contains reference to the original article, list of tags that can be key words or any other words that help to find it including their synonyms and abbreviations to facilitate search and grouping similar facts from different articles together. The card may also contain relevant references (if any). Tagging enables to assign the same fact to several different categories.


My current approach consists of two parts.

First, I make use of a Reference Management Software (in my case Papers), which allows me to highlight/annotate PDFs and to write notes for each PDF. It's easy to do a search which covers all the notes that I write, and even the internal text of the PDF. PDFs may be tagged or assigned to multiple categories/subfolders; they can also be exported with or without annotations.

Some examples of Reference Management Software (of which I have experience with 1-4):

  1. Papers
  2. Mendeley
  3. Zotero
  4. Sente
  5. Endnote

Second, I write a summary in a review article for each major topic that I am looking into seriously. I previously used Microsoft Word to do this, but have since switched to LaTeX/bibtex due to the convenience of being able to handle hundreds of references (or equations). The usefulness of a review article format is that it is easy to convert into a manuscript (keeping equations, references intact) when the time comes to write a journal paper. Furthermore, should I need to revisit a research area that I have not touched for many years, a review article allows me to quickly refresh my memory. Having all the relevant references in a single place also helps. A side benefit is that I sometimes send my review articles to collaborators to help them get up to speed.


I started a new strategy as follow: I am reading the article, taking notes, and then generating Q&A of the most relevant facts. I am using a flash card software with spaced repetition to help me review them. By doing so, I am no longer worried about forgetting things and more concentrated on understanding things. As I read more and more, I hope to reach the stage where only the recently published idea/concept are the ones which goes into my flash cards program. I know it's a tedious workflow, but I believe it will pay off soon.


There is no "best" way, everyone has their own way. I like to read the abstract and the discussion, and then the figures, and after that, the text. Its not something thats been tried and tested, you just develop your style over time reading hundreds of papers.

  • Thanks for sharing with me how to approach a new article paper, however how do you retrieve the new knowledge over a long period of time. Especially for someone who is new to the field facing a situation where almost every other fact is new.
    – Live
    Aug 12, 2012 at 20:48
  • 1
    Find a review on the subject area, they are the most comprehensive summaries of discoveries so far
    – gkadam
    Aug 13, 2012 at 0:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .