Tag Categories?

I use Zotero as my library of scientific papers. Of high importance is being able to search for a paper when I might not remember the authors/title/journal. Therefore, I try to add tags to each paper I download. Lately, I've found myself adding many tags based on specific categories (subject, major finding(s), material system, techniques used, etc.). That's a lot, and I often forget to add tags based on some of those categories, making subsequent searches less than thorough.

I think that having categories for these tags would be nice. So when one enters the metadata, they don't forget to add appropriate info. The 'tags' tab could then be a form that you create and fill out when downloading a new reference.

Another reason to have this is that currently the same tags will be used for different categories, and doing a search will not appropriately filter for those categories. For example, if I'm searching for a paper whose subject is a new method for "atomic force microscopy," I will also get papers that simply used that technique as a means to some other end. If I could additionally filter for that tag only in the "subject" category, the filter would be far more effective.

In a more radical change to Zotero, the left pane that has the folders could instead be different ways of filtering the tags. I am constantly unsure of which folder to place the new downloaded reference since they do not always fit into just one folder nicely. Later, folder navigation is useless to find the paper since I may have filed it under some other folder. If instead the folders or "collections" were simply categories of the tags and "subcategories" were the tags themselves, then the appropriate papers would always be in all the appropriate folders. The interface could still look the same, but there wouldn't be this rigid construct that makes it difficult to use. Instead it would be akin to a dynamic search. The user starts at the top level based on what s/he knows about the paper and converge on the paper of interest with simply a few clicks. A similar motif is used all the time in retail sites where various categories are always available to filter. Even Amazon.com does this.

I think that this could be an idea that Zotero could take to the bank, showing it's better than Mendeley or other similar programs. Accessing/finding the references quickly is still a headache for all these programs.
  • I think that you have misunderstood how collections work. One item can belong to multiple collections, so what you describe should be possible with collections already.
  • I agree an item belonging to multiple collections is similar, but locating an item would be more efficient if it was based on tags rather than dragging an item into multiple folders. Choosing a folder would also not be the same, since you could choose multiple 'folders' in different or the same categories, which would act as a filter to specify your search.

    In the near term, I'm more interested in simply having tag categories that could be searched in the advanced search option.
  • I doubt this is going to happen. Given the existence of collections, the added benefit is very small, the added work is significant because it would include re-working the internal db structure for tags as well as GUI issues.
    There are countless issues in Zotero that should be addressed that wouldn't essentially double existing functionality: batch editing, hierarchical items, a new html (and maybe pdf) annotation feature, more flexible note organization etc.
  • You might want to experiment with the recursive collections hidden option explained here : http://www.zotero.org/support/hidden_prefs
    Rather than just seeing the items placed in a collection, you see all the items in the various subcollections. I find that a very powerful way of finding papers I can only vaguely remember by drilling down through my hierarchical structure of collections. Often I will remember, for example, that I have used a text in teaching but not in which year of which course. Starting out in my teaching collection narrows things down and I can then take thigs down to successive sub-collections. For me that comes fairly close to filtering.
  • Thank you for your comments.

    To adamsmith, I would say that there's actually a large benefit to being able to find items in ones library if the current system is inadequate. It would certainly be a larger benefit than the list of desired functionalities you provided (although I'm not familiar with each of them). You may not have such an issue, but currently I often can't find the papers I'm looking for due to lack of an adequate search. I find that Googling for the paper is often a more efficient way to find a paper I've already read than using Zotero even with the current functionality. What's the use of a library you can't navigate?

    I also don't think that the coding would be significant. There could certainly be tags with no category. So if you haven't made categories, nothing would change. If you do make categories, they could show up as a list in the tags tab, and you can add tags under them just like you add authors.

    To clio_13 thank you for the suggestion. That would certainly help and be similar to what I was proposing in terms of replacing folders with tags.
  • I also don't think that the coding would be significant
    well, then just do it.
    The code is here: https://github.com/zotero

    Otherwise, I suggest, if you can't find your items you're using existing features wrong.
  • What makes this difficult to implement is that changing the data model to allow hierarchical structure for tags would require changes to Zotero, Zotero server, Zotero server API, Zotero website, and potentially other software using the API. These changes would have to be done in a way that maintains backward compatibility.

    However, there is a half-way solution that would be relatively straightforward to implement: grouping of tags based on prefix in the Zotero client. For example, if you had tags subject:chemistry, subject:physics, and subject:math, these would be combined as one tag category "subject" in the user interface. Tag categories could be shown in a separate section in the tag selector. The advantage of this approach is that it would only require changes in the tag selector code.

    Would this solve the problem presented in this feature request?
  • I see. I think that would be a good compromise. I assume the user would simply write the full string (e.g. "subject:chemistry") as the tag?

    It might not be too hard to eventually have the existing subjects listed in the tag tab and a user could click on one to add a tag under that category (which would be less work for the use).

    I'm actually not accustom to open programs. Is the person who suggests a feature change supposed to change the program his/herself as suggested by adamsmith?
  • I'm actually not accustom to open programs. Is the person who suggests a feature change supposed to change the program his/herself as suggested by adamsmith?
    Not per se. But I think adamsmith was pointing out that a) features don't happen by magic, but by work, b) this isn't the best use of limited resources (I personally think it's not a good idea, and would underline adamsmith's point that it's better to learn how to use the existing features better; they're powerful enough), and therefore, c) unless you want to write it yourself (which is possible) it ain't gonna happen.
  • edited January 26, 2012
    However, there is a half-way solution that would be relatively straightforward to implement: grouping of tags based on prefix in the Zotero client. For example, if you had tags subject:chemistry, subject:physics, and subject:math, these would be combined as one tag category "subject" in the user interface.
    FWIW, I think this is overkill. Tags are, by definition, lightweight categorizations, and I have a hard time seeing how them + collections don't cover user needs. I prefer to keep it simple, and consistent with tags as widely implemented outside Zotero.

    To go back to the OP, and this example:
    Another reason to have this is that currently the same tags will be used for different categories, and doing a search will not appropriately filter for those categories. For example, if I'm searching for a paper whose subject is a new method for "atomic force microscopy," I will also get papers that simply used that technique as a means to some other end. If I could additionally filter for that tag only in the "subject" category, the filter would be far more effective.
    My suggestion: if you want hierarchy, use collections. So maybe you have a top-level "methods" collection, and then subcollections for specific methods. I would use tags for subjects, but I sometimes create a specific collection if I have a tag that is the basis for a manuscript.

    Keep in mind, too, smart collections.
  • I'm actually not accustom to open programs. Is the person who suggests a feature change supposed to change the program his/herself as suggested by adamsmith?
    No, but if you're not going to, it's better to avoid suggesting that something would be easy. That's what adamsmith was responding to.
  • I see that the ability to assign tags into categories. The primary reason for me to not use tags at all is that they cannot be organized efficiently. A large group library that we use has hundreds of different tags. Since the tags take several screens worth of vertical space, I need to scroll back and forth to locate the ones that I want to use.

    Being able to browse the tags by category would be valuable for me. I am not in favor of hierarchical tags because they would add complexity.

    Collections do not work as a substitute for tags because you can only have one collection chosen at a time.
  • I agree with Waverlybrian about the usefulness of tags categories (or metatags), and I think it is a major thing, for people who need to use Zotero to code and sort articles somewhat seriously. And collections are not an effective substitute, as folders do not have the flexibility that tags have. That would mean for instance recreating a whole set of collections every time I work on a new project or a subcategory of the articles. Actually, with well crafted tags, that is collections that we would not need anymore.

    So far, I have been using the workaround of "category: detail" tags as suggested. It works, but 1- it gets clumsy after the second level. It becomes a chore to type and you cannot filter articles to "category: subcategory 1". 2- a typo and your article gets "lost" 3- reassigning tags between categories or more generally managing categories (like renaming them) gets clumsy as well. 4- I also agree with Waverlybrian that some sort of form, based on a customized list of tags' category, would be a heaven for me to code articles, and avoid missing categories altogether.

    It would be great to be able to vote when we think a certain feature is useful. In the meantime, we should be cautious before asserting that a request is unimportant or useless. I believe coding articles is a very basic and prominent need of researchers when it comes to documentation management.

    Thanks to all those who did and will improve Zotero. I wish I had the talent for that.
  • I agree with both Waverlybrian as well as Nuijel about the usefulness of tags categories (hierarchical, of the parent-child variety at least and, if possible, including in the design some of the user-defined functionalities outlined as a design proposal of sort by Waverlybrian).

    There was the beginning of a discussion on that front
    here:
    http://forums.zotero.org/discussion/700/equivalence-of-collections-and-tags/

    here:
    http://forums.zotero.org/discussion/79/hierarchical-tags/

    and here:
    http://forums.zotero.org/discussion/7168/how-do-people-organize-their-content

    Weighing in on the side that there should, perhaps, be resources devoted to that (and so being a promoter of some sort), I would argue that it might be in the business model, so to speak, of Zotero to stay upstream of tools like NVIVO, and enabling the strength of both to work well together. Zotero is a more natural place to do the bulk work of classifying tags, while NVIVO's ontology editing functions are more suited for finer grained tuning up.

    When one wants to see the whole side of a scientific field, involving many scientific debates, the knowledge of which being of import in the conducting of one's own research, than the hierarchical tagging can go a big way into making sense of a complex landscape.

    I've always believed that integrating the ontological classifications (the classified "level-of-realities" at which the scientific parties to a debate take place) would go a long way toward making sense of a scientific debate, thereby, perhaps and if we're lucky, lead to some form of Critical progress, to give this discussion a philosophy of science twist.

    I think an argument could be made that users will get more sophisticated over time and that we are in a "continuum of innovation", and ... [related Web 3.0 arguments]; and that tools like zotero will increasingly be relied upon to not only manage your researches but also to conduct your researches.

    In the threads referred to before, some people were laying out examples of tags that they thought referred to different "levels" of the reality of their fields ― while still being semantically the same ― and coming up with a) some kind of cases for the development of features of the type discussed here as well as b) sharing solutions for using the existing features to their full ontological potential.

    I wouldn't know whether or not some of the features discussed make a lot of coding/design sense in their approach/proposal, but they are usually given with some intentions to simplify the coding presumably needed by the feature requested/proposed.
Sign In or Register to comment.