The case for personal tags (long post alert)

heidib

In order for the power of Zotero to be leveraged as efficiently as possible, I would propose considering the types of metadata that might be attached to a document, and that the program's interface should easily fascilitate each type of metadata. Currently Zotero efficiently handles four types of meta-data (summary info, descriptive tags, notes, and relevance to a user's particular project). I propose a new, intuitive way to handle a fifth type of metadata ("personal tags").

Metadata type/ description	Function	variability/ general interest	entry/view/ searchability
Standard data about the document (author, publication date, abstract, etc)	Bibliographic and summary information for use in citations, reports	No variablity/high general interest; there should be one set of such data per document (eg it should not be customized for each user.)	Pulled in from catalog sources or extracted from PDF's / Metadata is separate tab entry / search on individual fields
Descriptive tags such as "intonation"	More precise than full text search on keywords, since there has been some human judgement applied to how relevant a term is to the article	Ideally no variability; standardized terms help keep related articles categorized together. But in practice there will likely be some variability eg "intonation", "prosody." Zotero's feature of letting you see all tags will help users use consistent keywords	Pulled in from catalog sources or extracted from PDF's / view tags in separate tab entry / Clickable to sort, very intuitive
Notes	Freeform comments or annotations on an article. Notes may be used to summarize article for a lit review or to paraphrase/ quote key points for inclusion in an article or paper	completely variable; low general interest; most notes will focus on aspects of a paper that is of interest to the note author or for a particular paper	Written by zotero user, also most attached documents fit into this category / can be searched
Relevance to a particular project	Keep together articles related to a class, research paper, dissertation, or topic of interest to the user	High variability / Low general interest; other students in your area might want to see your library of PhD references, teachers may want to share course materials with their students, but other than those types of cases no one really cares that you refered to an article in your second year semantics seminar.	Use folder system -- this works very well for this purpose, esp. as it allows items in the library to apply to more than one project / Fully intuitive and transparent search; click on folder name to see contents.
User's relationship to a particular article	Keep track of what you have read, skimmed, printed, kindled, plan to read, etc. Also, provide a way to dismiss an article as not useful or evaluate it as highly useful	Low variablity (most readers will want to keep track of the same types of things) / but no general interest.	Proposed: Add Personal Tag tab to main menu. By default, no personal tags set. Users can choose from preset tags or add their own / search should behave like regular Zotero tags

Because there is currently not obvious way to mark articles, for example, as already read, each new user must come up with their own system. They could make and "Unread" folder, for example, or add an "unread" note, or an unread tag such as "UNREAD" or "HB_UNREAD." While such options may work well for some people, for many users these options may the effect of weakening these modalities. For people going in with no preconceptions of the best way to handle this issue, they need to come up with a system by hit or miss (if I note an article as "READ", later will I remember if that is the past or imperitive tense of the verb? What if I forget and annotate some articles with "PRINTED" notes and others with "PRINTED" tags?

Furthermore, as Zotero heads off into the clouds, I think it is critical that the interface provide a distinction between annotations of general interest and those that are particular to the user. If this issue is not addressed, tags such as "HB_ADD_TO_KINDLE" will creep into the general library, reducing the signal to noise ratio of tags significantly.

Even custom "ranks" of articles are not likely to be of general interest OR they may be a misleading indicator. For example, I may rank very highly one of my supervisor's articles on which I may model my dissertation research. This shouldn't be generalized to mean this would be a useful article for everyone.

Personal tags that use a common vocabulary could be used to leverage a lot of power from Zotero. As I head into doctoral research, I can see that the same articles will come up frequently in library searches. It would be nice if Zotero could tell me if articles on a page had already been read or ranked before it adds them to my library so I don't waste my time on them again. Conversely, an option could let you mark results of a bibliographic search as "TO READ" before they even hit your library, enhancing the usability of the personal tag system.

Some proposed built-in personal tags (to start a discussion) :

is_read
is_skimmed
is_printed
is_on_kindle
is_on_ipod

to_be_read
to_be_printed
to_put_on_kindle
to_put_on_ipod

rank_very_relevant
rank_relevant
rank_not_relevant


Thanks for a great product!

-- Heidi

bdarcus

I think your argument that there's a distinction between different kinds of categorization makes some sense (though see below). But I think that observation actually suggests that there's no need to standardize what you all "personal tags." Just to underline the point, I would be annoyed to see all the options you list (I don't own either a kindle, or an ipod, for example, and I don't typically print articles).

I also think there's overlap between the personal and the social. For example, whether I've read something or not is certainly largely my business, and that metadata is really only for me. But it might have some larger relevance in aggregate. The same is even more true of relevance rankings.

alexuw

I've found myself thinking along similar lines as Heidi, but end up in a quandary. The power of tagging lies in its simplicity - the way tags are unstructured and open-ended. Obviously that means tags can quickly become a big mess, but when you're dealing only with your own this is less of a problem. But then you can imagine tags being that much more powerful ... if only ...

The problem is that making tag data more structured makes tags more complicated (especially data entry), less open-ended (requiring more forethought about how you'll use them), and introduces usages that some users don't want.

Heidi's proposal is dramatically simpler than the idea of hierarchical tags that has come up (see post linked below), and I wonder if something along those lines would be a happy medium. I probably wouldn't favor having any "built-in" predefined tags at all, just different classes of tag that could be shown and hidden at will (this already exists in one form - between auto-generated and normal tags, users could add their own e.g. "personal" tag type).

The data structure is already there but the required UI modifications would be pretty significant. And the additional complexity might be a mistake - tags work because they're dumb.

A couple posts of interest:

http://forums.zotero.org/discussion/79/hierarchical-tags/
http://forums.zotero.org/discussion/700/equivalence-of-collections-and-tags/

dstillman

http://forums.zotero.org/discussion/3544/change-the-type-of-tags/