The dissertation of @publicvoit on TagTrees, while interesting in its own right, also has a very interesting overview of historical #PIM #HCI research:
https://karl-voit.at/tagstore/downloads/Voit2012b.pdf#chapter.2
Oddly, though, even though it focuses on organizing personal information including files using free-form tags, and even talks about photo management and bookmark management as application areas, it doesn't mention Indecks cards (or other edge-notched cards), Lotus Agenda, or Flickr. It does mention del.icio.us once in passing on p. 67.
The TagTrees method he proposes doesn't rely on organizing the tags themselves in a tree (like Agenda). Rather, the #tags in a TagTrees path are intersected without any concern for their order; the tree view is just an "associative browsing" affordance.
@publicvoit Well, I haven't read *all* of it!
I think Agenda was relevant to the dissertation because (a) it organized items primarily with tags (called "categories", with initial tagging of each new item applied using text matching rules and implication rules) and (b) it was the program the term "PIM" was invented to describe.
I agree about the drawbacks of the cloud. As the wise man said, there is no cloud; there are only other people's computers. #TahoeLAFS is a solution to the resulting loss of autonomy, but it's not widely adopted.
I don't think Flickr and del.icio.us were "totally different from PIM in almost [every] aspect". Most people used them for managing their own photos and bookmarks, in precisely the way TagTrees was designed to be used; they didn't allow you to apply or even suggest tags to other people's photos or bookmarks. (On del you could bookmark someone else's page, of course, with your own tags.)
@radehi Yes, you're right. The "categories" were not categories in the usual sense. They'd qualify as tags. 👍
I disagree on flickr and del.icio.us since they are usually mentioned as examples of a https://en.wikipedia.org/wiki/Folksonomy where you tend to add as many tags as possible including synonyms and plural/singular forms to get a better chance of being retrieved by myself and others. For various reasons, this was not a viable approach for the technical implementation of tagstore (my research prototype).
@publicvoit With respect to Hypercard, the only things it has in common with Agenda are that it runs on a personal computer and that you can store data in it and search that data. Hypercard is a card-based graphical hypertext system embedding a high-level scripting language and a data entry screen system; Agenda is a faceted browser of text snippets that classifies them into a faceted tag hierarchy using a user-defined set of text matching rules and can't handle graphics.
@radehi Oh. https://en.wikipedia.org/wiki/Lotus_Agenda did not mention facets (which I personally do like!). I looked into adding some facetted approach into tagstore but failed to come up with a good concept for it.
@publicvoit In Agenda there's a "subcategory" relation, which is used for three things.
First, if A is a subcategory of B, then you have a logical implication: if an item has category A, it also has category B.
Second, in a table view, you can use B as a facet, with item values for that facet being A and B's other subcategories; this permits grouping items by B or adding a B column.
Third, you could optionally make B a "mutually exclusive" category so that an item can belong to at most one of its subcategories; this is especially useful if you want to use it for grouping or in order to tune your automatic classification rules. If you didn't do that, you might have items that were tagged with more than one of B's subtags, so in a table view you might end up with a cell with multiple values in it.
This simple model got more complicated in Agenda 2 when they added categories with associated values (for things like dates, money amounts, and temperatures.) Probably it would have made more sense in that paradigm to replace mutually exclusive subcategories with an enumerated data type.
@radehi Very interesting properties, thank you.
A tag hierarchy is quite common in many non-trivial tagging tools. I personally don't need it but I understand its purpose and value.
Table view: cool.
mutually exclusive tags: I've integrated a mutual exclusive feature to my #filetags project: https://github.com/novoid/filetags and I'm using this quite often.
It's nice to have an interaction who obviously has much knowledge about PIM research and tools! 👍
@publicvoit Likewise!
If you're interested in seeing how #LotusAgenda did things, there's a [slow-paced #demo on YouTube starting at 8'16"](https://youtu.be/GsJzRv-UDUM?t=496) The previous several minutes are mostly an explanation.
Like, *very* slow paced; he [finally adds an item at 17'28"](https://youtu.be/GsJzRv-UDUM?t=1048). Later he demonstrates [auto-tagging based on text content](https://youtu.be/GsJzRv-UDUM?t=1318) and [at 37'36"](https://youtu.be/GsJzRv-UDUM?t=2256).
A thing I'd forgotten is that when you're adding an item in a view with a column for some category C, you can type a new value D into that column to create a new subcategory D within C.
(He also demonstrates its features only peripherally related to #tagging, like [an item being sorted on entry](https://youtu.be/GsJzRv-UDUM?t=1239), [adding a note to an item](https://youtu.be/GsJzRv-UDUM?t=1095), and [setting a value in a numerical category](https://youtu.be/GsJzRv-UDUM?t=2152).)
@publicvoit the filetags idea of storing tags by renaming files is a new one for me; previously I've used either metadata-file-per-tag (`echo heartshapedbox.mp3 >> rock.playlist`) or metadata-file-per-file (`echo rock >> heartshapedbox.mp3.tags`), with convenient scripts for adding and querying with tags thus represented. Tab-completion and mutual exclusion is definitely a plus.
@radehi Furthermore, I would not recommend using this kind of tagging strategy for personal items as described also on https://karl-voit.at/2022/01/29/How-to-Use-Tags/ independent of technical limitations of having "too many" tags. This is not based on research (yet) but on my personal experience. YMMV.
@publicvoit Oh, yeah, people do tagspam sometimes on Flickr! And that's partly a social thing (though sometimes it can be useful, as you say, for when you forget your tag names yourself). A controlled vocabulary takes some effort to manage, and existing software doesn't do very well at it.
@publicvoit Still, I do disagree with your implicit equation of folksonomies with tagspamming.
@radehi Fair enough. Do you have counter examples of a (public) folksonomy where there is no "tagspamming"?
@publicvoit No, I'm pretty sure you'll find tagspamming anywhere there's a public folksonomy, just as you'll find unsolicited sales calls in any medium that permits contacting strangers. But that doesn't mean unsolicited sales calls are the same thing as telecommunications!
@radehi Somebody who actually read my thesis. 🙇 👋
I didn't know Lotus Agenda but the descriptions sounds like many other tools such as Hypercard.
Other things didn't get mentioned because the thesis is focusing on PIM and not collaborative tagging which is totally different in almost any aspect.
Furthermore, I personally don't like the cloud for personal data: https://karl-voit.at/cloud/ and https://karl-voit.at/cloud-data-conditions/
HTH