Scientific Researchers and Web 2.0: Social Not Working? forum: topic
This is a public forum
Organizing references by tags or search (or both)?
Maxine Clarke
Monday, 23 February 2009 17:35 UTC
David Crotty, of Cold Spring Harbor Protocols and Bench Marks blog, writes a downbeat post about online social tagging and why in his opinion it isn’t working. Social bookmarking sites, in which you organize your references by tags, are tedious to use, he says, as well as adding bureacracy (writing summaries). He is keener on a full-text search system, citing Papers (which as far as I am aware is only a Mac application, not for use on a PC).
An opposite view, to which David is in effect responding in his own post, is that of William Gunn, who thinks that a system of expert reviewers tagging and recommending papers is the “killer app” of science 2.0.
I myself tend to rely on search rather than tags when looking things up, but I am not a researcher and, in the line of my job, I am constantly looking things up in different “universes” of literature. If I were a researcher, I am sure I’d have a reference library and I’m sure I’d find it useful to tag and annotate it. My experience of trying this out on Connotea has been very positive, and there are other similar services. I would not be happy depending on search for everything (especially if I had to go out and buy a Mac and learn a new operating system to do it!).
I’m interested to know what Nature Network users do, and think, about this question.
-
Replies
Jump to resultsResults
-
Tagging puts the onus on the writer. This is why we have computers. It should be automatic… hence search. Search search search.
-
I agree for general use, but for the specific purpose of one’s reference libraries, though?
-
So, for my two cents — I tend to side with Maxine on this one, at least where the usability of personal tagging comes in. As I understand it, “searching” means using someone else’s keywords for retrieval — hence, when using Pubmed, I also must use MeSH to navigate successfully. The same exists for any host service/database providing a search function: there are terms to use and when you don’t use them, the search fails or is extremely frustrating. To experts, this is one of those forgotten traumas. To new scholars, it is an ever-present nightmare. Enter tagging: with an “expert system” or the one being brainstormed in Citations in Science then networks of related articles with or without summarized reviews would lend another level of information about any particular piece of research (personally, I think Thomas’s idea of a quick tag system based on reader’s perceived relationships is great for fast commentary; the option of in-depth response is good, too, when one has more thinking time). So, now, there are 2 levels of search: something like “key terms” and something like “connections” with a peer review option. In both, the user still has to learn someone else’s categorization of information. This is okay — it’s part of becoming a disciplinary expert. It’s not so useful for the naturally inter-disciplinary professions such as GPs or comm/sci disorders where multiple specialities encourage both specific and generalist expertise. Here is where a personal tagging system is still important: at some point, the individual needs to be able to say how a particular bit of information fits into their particular schema. Social tag sites — delicious, 2collab, citulike — all have a level of crowd-sourced tag suggestions that a user can select (and which are informative in and of themselves) or the option to create idiosyncratic options. As a user of the scientific literature, I value/need many perspectives, but ultimately also want to exercise personal control over the information I find.
-
The trick with tagging is to remember to tag! And tag sensibly.
Maybe it takes a different mindset?
-
Yes, the only one of these reference sharing/tagging services I have used is Connotea, and contrary to what David writes in his post it is instant to add a reference to your library (using the bookmarklet). And as Mickey writes, in the window you see a suggested set of tags, or you can add in new ones, the choice is yours. You can annotate if you want, not if you don’t. You also have the option of just seeing your own tags in a list, or everyone’s, or everyone who has used that tag, etc – again, all quick. I don’t mean to sound like an advert, but if I were working in one field of research, I’d find a service like this very useful to essential. I would also use the saved RSS search-term function on PubMed too (if it is now working as it was intended to).
Richard, if you just depend on search, don’t you risk missing stuff that the search engine itself doesn’t index, or that doesn’t get pulled up by your own choice of search terms? I think I’ve a lot to learn about all this. -
You mentioned Connotea in the text of this discussion, but you replied Richard in terms of specific purpose of on’s reference libraries. In my opinion these are different situations and lead to different answers.
For personal use, I prefer category to tags. Category in my mind is much more stable than tags. I may tag the same item differently at different times.
However, if my items are also shared with other and desired to share efficiently, tags are more favorable than category because the standards of the latter vary more from person to person.
-
Well, Andrew, I think that Connotea operates only in tags, not categories, and that is the only reference management software I have tried.
I use categories on my blogs, but agree I don’t think they would be as useful for reference management. Categories on Nautilus have come in useful on many an occasion, though.
-
I tend to prefer text search, but like Maxine, I’m typically using quite a range of literature with little overlap (I’m a student). I’ve also tried Connotea, but I now use Zotero (zotero.org), which saves a reference with just one click. I keep references in collections that I define beforehand—typically, one collection per essay.
Would tagging be easier if the computer did it for us, by word trawling? E.g. if it mentions a species more than a couple of times, tag the paper with that species, or if it uses a particular family of words, tag it with the relevant topic? Of course, this would probably lead to us not doing any tagging ourselves, which rather goes against the idea of having human-definable tags.
Social tagging seems flawed to me because different people will categorise things in different ways: e.g. as a student, I may tag a paper “conservation”, but this is clearly unhelpful for a conservation biologist.
-
First, I don’t see why a “category” can not be a “tag”.
Second, I only essentially function with suggested tags if I do bother at all. The blank field gives me writer’s block – it’s the same for the blog, by the way, unless I consciously want to make it possible for someone to find my article or blog post with a particular keyword that is not present in the title or abstract.
I like the concept of tags that are generated for and then selected/approved by the individual because they speak to that person. Over a lot of people, there will be some consensus, but outliers are still tolerated.
There is a very thoughtful piece at Disease Models and Mechanisms that just appeared on a similar issue. One phrase that I liked and that stuck in my mind, because I don’t agree fully with all the assertions, was
The crowd is post-filtering the options; their aggregate actions reveal the latent content about the quality…
This is a little bit like what F1000 is trying to do (quality assessment), but wouldn’t it be interesting if the publishers (or perhaps PubMed via its LinkOut feature) made available the statistics on the number of people who accessed the full-length article? That might be a measure of how the (eg. MeSH) tags stack up as well, because people who are actually interested in reading it, are finding the article in question.
But personally, I mostly rely on search. I’m comfortable with searching for keywords, as I’ve used searches for about 20 years now. But I do miss things that I only see later.
-
In general, I find the philosophy behind organizing through tags to be tedious and inefficient. It requires you to know, in advance, everything you’re going to get out of a paper so you can tag it when you first get hold of it. If not, you have to hope you can find it again later on down the line and re-tag it, and re-tag it as your research priorities change. Why not just skip all that work and be able to find what you’re looking for with a quick search? Programs like Papers allow for full text searching of the pdf’s you deposit.
There’s a good debate on the sorts of programs I prefer, referred to as “everything buckets” going on here:
http://al3x.net/2009/01/31/against-everything-buckets.html
And a follow-up listing some rebuttals here:
http://al3x.net/2009/02/24/why-no-comments-more-everything-buckets.htmlAlthough I’d take a more silo-ed approach, using one bucket for music, one for photos, one for video, one for science papers, etc.
Maxine, as a Windows-user, perhaps the nearest equivalent would be iTunes. Note how you just dump in songs and the program sorts them for you, and you can then re-find them through a variety of means. You can also further organize to your heart’s desire, through further tags or by folders or playlists.
Maybe I’m not using Connotea correctly, but I have to tag each article I try to add there, although when I start typing in tags, it does suggest ones that I’ve already used in my library that start with the same letters. Not quite the same as suggesting tags ad hoc.
Results
-
