I use the online tool citeulike to manage my bibliography. The main tool to organize the library is tags (functionally equivalent to tags here on this site). One problem I seem to be having is I do a very poor job of initially choosing tags, so I need to continually re-tag my library to keep it organized.

Is there any advice on choosing an intitial set of tags? Or will it be necessary for me to continually maintain my library to keep it organized/updated as much as I would like?

  • I am not sure what answer would satisfy you. People have different conventions when it comes to tagging. What is your particular problem with your current tagging scheme? (And why do you need to retag things?) Commented Feb 20, 2012 at 19:40
  • @PiotrMigdal, I am aiming for advice about prospectively identifying tags or tag-systems that would be useful (beyond say key-words for articles). I need to re-tag things because I don't always know ahead of time what specific content of articles I would like tagged. For instance, I might say tomorrow I want to identify all of the articles in my library that have used cross-sectional analysis (versus longitudinal) for a specific project I have in mind. I did not make such tags initially, so I need to go back and re-tag if I want this information in the future.
    – Andy W
    Commented Feb 20, 2012 at 22:33
  • I'm not sure what answer would satisfy me either. I figured this would be a good community to ask if anyone has developed any more formal system (I certainly haven't given that much thought to it). Although I doubt a perfect example exists that would fit my need, any advice about "tagging conventions" and why they are useful for any particular case I suspect would be insightful.
    – Andy W
    Commented Feb 20, 2012 at 22:35
  • How about using multiple tags, some very general (e.g. sequence analysis) and some exactly related to your projects (with their names)? Then, after some time, you can search either for its general content (it was about X) or by your former project (I used it for Y). Does it seems OK? (I like tags and tagging but I cannot say much besides that redundancy is often usefull; e.g. for my delicious account I use typically 5 tags per webpage). Commented Feb 20, 2012 at 22:59
  • @PiotrMigdal, that seems similar to what I am doing now, it doesn't help with the prospective tagging part though (in essence I don't do a good job of foreseeing future projects). It may be that I should expect to have to go back and re-tag (hence I should spend effort making a system that I can re-tag articles in a fairly trivial way).
    – Andy W
    Commented Feb 20, 2012 at 23:44

4 Answers 4


Tagging is only useful if you use it with discipline. Look at Wikipedia or Stackexchange, most articles can be determined by 5-6 categories (science - math - geometry - euclidean geometry - metric). Stackexchange has a max. of 5 tags, often you only see 1-2 on questions, which is often pointless, as those tags will appear often in the question/paper title and abstract too. Wasted time.

So, if you decide to create a tagging system, use at least 4-6 tags depending on how interdisciplinary and specialized your paper/link collection is.

Also consider to not only tag by topic but also kind (review, letter, peer-reviewed, experimental results, theoretical analysis, explanation of new measurement method, meta-discussion...), year, personal rating (very interesting article you learned a lot from and should read again from time to time), rarely/often/top cited, new theory/model, strongly discussed in the research community

A last note. I use myself Copernic Desktop Search as a supplementary tool, I download all papers of possible interest (disk space is cheap ;) ), papers I read, will read or maybe will never take a look at. The point is that Desktop Search software often has more powerful search operators and sorting mechanisms than Google Scholar & Co. If you know how to use them, you can save a lot reading and tagging time or tagging at all. You know, if you are smart in using Google & search operators and have a good vocabulary, you don't have to ask a lot questions on internet boards.


Don't use tagging for creating a pure thematic and linear directory structure, if finding again your papers or bits of information can be done by learning a good Desktop Search software. Use your tags in a personal way and remember, the point is not to structure your bibliography like a folder directory for categorized files, the point is to find again the bits of knowledge and most memorable papers, which will rather look like a strongly interconnected nonlinear tag cloud. If you look how people tag sites on del.ic.ious, often only 2 or 3 tags, sometimes using up to 10 pure thematic redundant tags, they are doing it imho wrong and waste a lot of time.

  • Given all of the suggestions to utilize text searches more effectively I might have to expand my tool-box. Thank you for the insight.
    – Andy W
    Commented Mar 1, 2012 at 17:27

I don't use citeulike myself, but I think it's important to make a difference between tags and keywords. Indeed, keywords are usually already included in the paper or at least in the abstract, so you don't need to retag them with duplicate information. Instead, it's better to use tags to give some personal context, such as why did you read this paper in the first place, or which general idea can you connect it to, or for which of your papers you used it, etc. In this case, there is no "wrong" or "right" tag, just some facts.

  • The comment about personal context jives well with my experience. I may not expect to be able to prospectively have all the tags I may at sometime want, but I find in my experience I can find what articles I am looking for if I have some sort of aid to jog my memory about the content of the article. For some reason author names sometimes help me with that connection, but I could likely work on utilizing tags to the same extent.
    – Andy W
    Commented Feb 20, 2012 at 22:44

The best solution of my is Papers from mekentosj. I should also mention that using tags is not an efficient solution for me; at the tagging moment I could not think about relevant tags and later I would not be able to find the paper. In my system I organized papers in collections (like folders) and search for them using keywords (like one would on in Web of Science) or authors or any other usual field.

  • 2
    I don't see how this answers my question at all, care to elaborate?
    – Andy W
    Commented Feb 26, 2012 at 12:26
  • You can organize your papers into folders, collections, etc. Then you can search for them using tags you add to the notes of each paper Commented Feb 26, 2012 at 14:34
  • The same fundamental problem exists whether I use tags or folders (I could create tags to function equivalently to that of folders). I would need to know what folders/collections to create to begin with.
    – Andy W
    Commented Feb 26, 2012 at 14:49
  • Working with Papers2 I do not care about tags since I do not specify or create any. I create folders for the sake of some organization, and later, when I try to find something, I go to the "search" box and type relevant keywords related with the things I am trying to find. It works usually, at least for me. Commented Feb 26, 2012 at 15:58
  • So are you saying that searching through documents is a much better solution than relying on tags? I think that is reasonable (one of the comments to my original question made the same suggestion). You should edit your answer to include that sentiment.
    – Andy W
    Commented Feb 26, 2012 at 16:07

One of the great things about CiteULike is that you can see what papers you have in common with other people, and what they're tagging them as. That can give you a good clue when you're starting tagging.

However, whatever you choose, it will evolve as you aggregate a larger and larger library of references, so don't be afraid to keep it flexible.

In the end, your tagging system will be unique to you, because your interests are unique. For example, in my library, I've got a medical-epidemiology tag, because that covers around 1% of my library. If I had no such papers, it would be a useless tag. If I were doing medical epidemiology research, it would also be a useless tag, because I'd have to apply it to 90+% of my library.

Learn by your own experience: how do you use tags? Which tags have helped you find things? Which tags have you used to try to find things, that turned out to be dead ends?

You must log in to answer this question.

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