Zettelkasten Forum


The Difference Between Good and Bad Tags

Comments

  • You don't say what tags you would in fact assign to your note on insulin sensitivity and carbohydrate intake. It seems like it would be useful if you'd show us what you would actually do rather than saying what one shouldn't do.

  • edited September 29

    Could you give me the German Words you would use for “topic” and “ object”.

    In my Mind a note is about one topic and i think you called that the object of the note. I’m confused.

    Kind regards,
    Dick Barends

  • edited September 30

    I'm also confused by the distinction you are making between "topic" and "object". From context it seems you are using "topic" to refer to the broad interpretation of a topic, and "object" to refer to the narrow interpretation of a topic, but I'm not sure.

    I've been trying to think of "object" in different senses to get clear on what you're getting at: physical objects, the class objects of object-oriented programming, the grammatical object of a sentence, objectives/goals -- but I'm not sure where your use of the word fits.

    The overall message seems to be "define your tags narrowly rather than broadly", but I'm not sure.

  • Ditto, re the confusion. It is not clear to me why #diet applies to 'metabolic flexibility' more precisely than it applies to 'insulin sensitivity,' probably because I am not so familiar with the subject of the example. An example on a more general subject might be helpful for the non-nutrition-minded folks among us. ;)

  • Thanks for the critics. You are right. I implicitly presupposed some metaphysics. I'll clarify in more extensive article. For the questions:

    Tags

    Tags on insulin sensitivity: #insulinsensitivity #metabolicflexibility #glucosemetabolism. These are objects in a manner that they are entities. Things in the world.

    German

    "Thematisch" for "topic"; "ontologisch" for "object"

    Difference between topic and object

    A topic might be a somewhat arbitrary umbrella. If I write about nutrition I can include topics like environment issues, ethics, etc. But I could also exclude them -- arbitrarily.

    An object is some empirical or theoretical entity that is more rigidly to accept. Diet is the set of behaviours that govern the intake of various nutrients. So it would apply more to the act of eating, habits and so on. But the ethics are not part of diet as the set of behaviours but rather the broader context and context is always a separate thing.

    Quote from the article:

    But I’d use the tag for a note on metabolic flexibility as overarching concept that connects diet and fasting.

    I buried the addition in the footnotes:

    Diet and fasting are two of six categories in my work on lifestyle. These are not topics but principles and therefore abstract entities in my work.

    So in summary: Topics are somewhat arbitrary. Entities are empirical or theoretical things in the world.

  • edited October 1

    Thanks :) Even though I don't speak German, the addition of the German words helps a lot -- since they seem to share roots with "Theme" and "Ontology". This helps narrow down the (ridiculous number of) possible meanings and usages of "topic" and "object". The article makes sense in terms of a thematic topic and an ontological object.

  • Earlier this morning I was reading The Difference Between Good and Bad Tags and the discussion of topics versus objects got me thinking about semantics on my website in general.

    People often ask why WordPress has both a Category and a Tag functionality, and to some extent it would seem to be just for this thing--differentiating between topics and objects--or at least it's how I have used it and perceived others doing so as well. (Incidentally from a functionality perspective categories in the WordPress taxonomy also have a hierarchy while tags do not.) I find that I don't always do a great job at differentiating between them nor do I do so cleanly every time. Typically it's more apparent when I go searching for something and have a difficult time in finding it as a result. Usually the problem is getting back too many results instead of a smaller desired subset. In some sense I also look at categories as things which might be more interesting for others to subscribe to or follow via RSS from my site, though I also have RSS feeds for tags as well as for post types/kinds as well.

    I also find that I have a subtle differentiation using singular versus plural tags which I think I'm generally using to differentiate between the idea of "mine" versus "others". Thus the (singular) tag for "commonplace book" should be a reference to my particular commonplace book versus the (plural) tag "commonplace books" which I use to reference either the generic idea or the specific commonplace books of others. Sadly I don't think I apply this "rule" consistently either, but hope to do so in the future.

    I've also been playing around with some more technical tags like math.NT (standing for number theory), following the lead of arXiv.org. While I would generally have used a tag "number theory", I've been toying around with the idea of using the math.XX format for more technical related research on my site and the more human readable "number theory" for the more generic popular press related material. I still have some more playing around with the idea to see what shakes out. I've noticed in passing that Terence Tao uses these same designations on his site, but he does them at the category level rather than the tag level.

    Now that I'm several years into such a system, I should probably spend some time going back and broadening out the topic categories (I arbitrarily attempt to keep the list small--in part for public display/vanity reasons, but it's relatively easy to limit what shows to the public in my category list view.) Then I ought to do a bit of clean up within the tags themselves which have gotten unwieldy and often have spelling mistakes which cause searches to potentially fail. I also find that some of my auto-tagging processes by importing tags from the original sources' pages could be cleaned up as well, though those are generally stored in a different location on my website, so it's not as big a deal to me.

    Naturally I find myself also thinking about the ontogeny/phylogeny problems of how I do these things versus how others at large do them as well, so feel free to chime in with your ideas, especially if you take tags/categories for your commonplace book/website seriously. I'd like to ultimately circle back around on this with regard to the more generic tagging done from a web-standards perspective within the IndieWeb and Microformats communities. I notice almost immediately that the "tag" and "category" pages on the IndieWeb wiki redirect to the same page yet there are various microformats including u-tag-of and u-category which are related but have slightly different meanings on first blush. (There is in fact an example on the IndieWeb "tag" page which includes both of these classes neither of which seems to be counter-documented at the Microformats site.) I should also dig around to see what Kevin Marks or the crew at Technorati must surely have written a decade or more ago on the topic.

    Original reply posted at https://boffosocko.com/2018/10/03/some-ideas-about-tags-categories-and-metadata-for-online-commonplace-books-and-search/

  • Hi!

    I enjoyed this post and have also been thinking about tags recently. Picking good tags is difficult and so is staying consistent.

    One thing I've thought about recently, is "namespacing" tags. For example #meeting #people/john #people/jane could tag a description of a meeting that I had with John and Jane. Another example: #companies/amazon could tag something I learned about Amazon or a journal entry where I discussed a package or order related to online shopping. This has two advantages: (1) it avoids collisions (e.g. if you knew a person and company with the same name), and (2) with a proper search you can easily see all people, companies etc that have been tagged.

    I'm wondering what you think about this and whether you'd consider it a good approach or perhaps even an antipattern.

  • @ms9 I experimented with things like that (with the flickr style of separating parts with a colon) because it looked clever and fun at first, but I found this doesn't scale. It creates the same problems like directory hierarchies, where topics come up in different contexts, like in #productivity/reading and #hobby/reading, or #webdesign/layout and #print/layout. I think it can work for some limited domains where you don't want to clutter the global namespace, e.g. for #people/john, but I would use that for meeting notes that are going to be deleted some day, not for things that last. You get similar results for topics by searching for the intersection of sets of #productivity AND #reading; you do lose the information about which comes first, but that hardly matters in practice.

    Author at Zettelkasten.de • http://christiantietze.de/

  • @ms9 said:
    One thing I've thought about recently, is "namespacing" tags. For example #meeting #people/john #people/jane could tag a description of a meeting that I had with John and Jane. Another example: #companies/amazon could tag something I learned about Amazon or a journal entry where I discussed a package or order related to online shopping. This has two advantages: (1) it avoids collisions (e.g. if you knew a person and company with the same name), and (2) with a proper search you can easily see all people, companies etc that have been tagged.

    I too have found some functionality with that method, but am also worried about the unknowns of how it will scale and what functionality I might be losing. The following are what I think are benefits, but I'm curious to get anyone else's perspective:

    There are certain tags that overlap areas of life—like #contacts, #idea, #article, #people, #meetings. I work in the entertainment industry, but I also have a business (XYZ) in the fitness industry. So it makes sense to me to have tags like: #ENT.idea, #XYZ.idea, #ENT.people, #XYZ.people.
    — That way when I search: "people," I'll already have results sorted by ENT and XYZ
    — That way when I only search: #ent. I'll have all #ENT results already sorted for me. It'll look like #ENT.people, #ENT.idea, etcetera.
    — The benefit is that I don't have to remember all the tags related to #ENT. That means I could search: #ent and happily have the results show me long-forgotten tags like #ENT.projects, #ENT.movies, or #ENT.writing
    — And I don't lose functionality if I just search: "projects" because I'll still get all the #ENT.projects, along with the #XYZ.projects, along with random notes I made that I just tagged with #projects

    What are the drawbacks of an approach like that for what I'm calling 'overlapping tags'?

  • I've started using prefixing for topic names that cover a broad area that it's useful to distinguish from other uses of the same term. For example, wb- for worldbuilding related, cl- for conlanging related (actually covers anything linguistical), pg- for procedural-generation related programming, and so on.

    "cl-argument" is about a grammatical argument (say, the indirect object of a verb) rather than a philosophical, logical, or programming argument. "wb-history" is about generating a fictitious history, while "history" is about actual historical events. "pg-mountains" is about ways to generate mountains procedurally, whereas "mountains" would be about the real world.

    They frequently reference each other, but they are distinct topics, and namespacing the tags and topic names helps.

    (ack, new browser. Needed to tell it that "worldbuilding", "conlanging", "procedurally" and "namespacing" are too actual words =^p )

  • I'm also confused why topic tags are a bad thing in some cases. In the example in the blog post, structure notes are used to create a table of contents, but I use tags to do the same, and it's much faster than managing a structure note. For example, if I have a tag #project3 and I search on that that, I get something like this:

    20180108 worked with dave to develop deck
    20180106 met with sally to bring in design elements
    20180105 handed off requirements to development team

    and so one. So by searching (or even saving the search for the tag), I in fact create a very nice log of events.

    I'm unsure why I'd want to spend the extra effort to create a table of contents for this kind of thing. Now I do have an inventory file, which keeps track of everything I own, and there are links to other files for extra information, and this makes sense.

    Can someone possibly show me a negative for my #project3 approach?

  • @nickmilo22 said:

    There are certain tags that overlap areas of life—like #contacts, #idea, #article, #people, #meetings. I work in the entertainment industry, but I also have a business (XYZ) in the fitness industry. So it makes sense to me to have tags like: #ENT.idea, #XYZ.idea, #ENT.people, #XYZ.people.
    — That way when I search: "people," I'll already have results sorted by ENT and XYZ
    — That way when I only search: #ent. I'll have all #ENT results already sorted for me. It'll look like #ENT.people, #ENT.idea, etcetera.
    — The benefit is that I don't have to remember all the tags related to #ENT. That means I could search: #ent and happily have the results show me long-forgotten tags like #ENT.projects, #ENT.movies, or #ENT.writing
    — And I don't lose functionality if I just search: "projects" because I'll still get all the #ENT.projects, along with the #XYZ.projects, along with random notes I made that I just tagged with #projects

    The same works using the power of set algebra:

    1. Replace #ENT.idea with #ENT #idea and #XYZ.idea with #XYZ #idea
    2. Search for idea or #idea and you will get the union of what used to be XYZ.idea and ENT.idea plus all ideas that are not part of either.
    3. Search for #ENT and you get all the tags related to entertainment

    There's no loss of functionality. My argument so far would be: Occam's Razor. Try not to introduce clever technical solutions when the existing ways to do things yield the same results. Because a drawback of #ENT.idea is, you won't get these when you search for all #ideas. Which you might want, because it is an idea, after all. So you could end up mass-tagging all #ENT.idea with #idea, too, once you run into this scenario. (That's what happened to me.)

    With apps that can do more sophisticated searches, you can write #ENT #idea NOT #XYZ and get the intersection of #ENT and #idea only, which equals the old #ENT.idea search.

    Author at Zettelkasten.de • http://christiantietze.de/

Sign In or Register to comment.