Suggestion for an improved tag management view.

One feature that keeps my from fully embracing DevonThink as my primary personal knowledge management system is the lack of an adequate tag instantiation/management system.

The only way to instantiate a tag in the present version of DT is to add it to a specific artifact/document, and then, after the fact, arrange that tag into the tag hierarchy. However, I maintain a fairly complex hierarchy of tags, and I want the freedom to think about them ‘abstractly’ – as divorced from specific artifacts. I find it a useful planning tool to create a tag hierarchy all-at-once, anticipating future needs – rather than to build a tag hierarchy artifact-by-artifact. I bet I’m not alone in this.

I have tried workarounds, but even these are frustrating. My latest bright-idea workaround (lasted five minutes) was to create a document that existed solely for the purpose of planning and managing my tag hierarchy. I thought: anytime I wanted to add a tag to my hierarchy independent (for the moment) of a artifact to attach it to, I’d just add it to this one giant, overly-tagged document – and then quickly arrange that tag into the hierarchy in the opened tag revealer to the left. This sounded great until I tried to add a gazillion tokens quickly to that teensy little bar. Hint: doesn’t work very well. This is not my quick tag management solution.


I think the ability to instantiate tags into the tag hierarchy independent of artifacts – as part of an improved tag management view – would really be a great addition to a future version of DevonThink.

It’s a powerful tool with a lot of potential, and I think this would really push it over the edge for me, as a user. Thanks!

I am sure this is a very excellent suggestion and technique. But, I wonder if others, like me, have trouble visualizing what the experience would be. I’m not grasping what it means to sit at the keyboard and “instantiate tags into the tag hierarchy independent of artifacts” and what “an improved tag management view” looks like.

Could you post a mockup or some way to help us understand what you’re seeing when you imagine this feature? Pictures, 1000 words, etc. :slight_smile:

Ack, sorry. That was an incredibly wordy post; feeling silly now.

I’ll try again. I want a tag management view where I have the option to simply “create new tag” … and then take that new tag and drag it into the tag hierarchy.

The only way to make a new tag, currently, is to use the tag bar at the bottom of every artifact. This means you can’t really plan your tag hierarchy ahead of time; you just have to make up tags on the fly, as you find use for them, when adding specific documents/artifacts/etc. into your database. I find it limiting for this reason. I want a better tag planning view, essentially.

Should I just delete my original post and say that? :wink:

Gotcha. You want to make tag groups before you use the tags. So, tags are groups, and Data > New > Group (or the equivalent in the toolbar or the contextual menu) is virtually the same as Data > New > Tag would be.

(Please leave the original post – it’s important advice.)

For the most part, tags are just Groups. If you want to create a tag hierarchy in advance, just create new groups inside the Tag Group, or create new groups anywhere and drop them into the Tags group. They will become regular tags once they are under the Tags group.

You can even organize tag hierarchies inside the Tag group by selecting a Tag, showing the Info pane for a tag and checking ‘Exclude from Tagging’ in the pane.

I understand you probably added the forward slash to differentiate a Tag from content data but I still advise against using characters generally reserved for UNIX (or other operating systems). It may seem a minor thing but it could cause issues when querying the location of a file and the first slash would have to be escaped in shell scripting.

Just a suggestion.

As korm and Greg have suggested, Tags Groups are just that. You want rapid-fire Tag creation? Select the Tag Group and press Command-Shift-N, type the Tag, press Enter, press Command-Shift-N,… (No need to reselect the top level group.)
If you want to make a hierarchy, do the same but after creating the top level Tag, press Command-O to open the Tag Group in a new window. Then repeat the same procedure as before, opening any Group you want as subhierarchies. (Command-W to close a window and return to parent Groups or the top level of Tags.)
Repeat until bored. 8)

Also, note that instead of relying so heavily on Tags, consider whether you really need them (especially large numbers of them). Though I come from a very Tags-based background previously (Ironic Software), I have come to realize that many times Tags are used to compensate for poor search capabilities. I find I am using Tags far less often with DEVONthink because its search functions are so powerful. So think deeply about why you’re Tagging before you commit too much time on how you’re Tagging.

Just a comment on the issue of ‘too many tags’…

I’ve recently started using the scheme proposed by J Ostwald in his blog (I think he also posts on these forums).

Basically, as far as I understand it, he suggests that you should invert the ‘normal’ method of using groups to hold documents and tags to categorise them.

He suggests it makes much better use of the DTP AI to hold your sources in tags, and to have groups represent the relevant topics of your research. Then when you annotate the sources - in small chunks (‘one thought, one card’) - See Also and Classify are able to propose which topic(s) the note should be replicated to. The point is that the AI works much better with groups than with tags, so use it with the things that matter – the topics, not their source.

So, for my research (Wartime Chester), I have top level tags representing Primary and Secondary Sources, then the next level for type (Newspapers, Minutes, etc) and finally the actual source tags (Chester Chronicle). I have a group hierarchy (‘Topics’) with many sub-groups to reflect my areas of interest (Court Cases / Crime / Domestic Burglary).

The workflow is then, import the newspaper clipping into ‘Chester Chronicle’, annotate it in a markdown note, then use Classify to propose suitable topics. I find it easier to type the topics into the tag bar, rather than drag and drop, but that’s just a preference.

You need to be careful with the properties of groups and tags though (shift-cmd-i). For example, you don’t want the Source tags appearing in Classify, so they need to excluded from Classification (but not from Tagging or See Also). On the other hand, you want the lower level of topics to be visible to both Tagging and Classification, but you probably don’t need higher levels – e.g. in my example above, it’s confusing and unnecessary to see Topics and Court Cases in the tag bar and Classify pane, so these are also excluded. The advantage of getting this aspect right, combined with the ‘short note’ philosophy, is that there’s less scope for the tag bar and Classification panels to be confused with huge numbers of irrelevant hits.

I’ve been using this scheme for about a month and so far it’s working very well. The benefit is that you’re getting help with the classification of notes, which you don’t get (as much) if you’re relying on tagging. It’s clearly not suitable for everybody, but it might be worth looking at for some. If you are interested, please go to J Ostwald’s site and read the details – it’s a very useful and comprehensive discussion of his research methods and I’m grateful to him for taking the time to document them.

Sorry for the long post, (and I’m sure that it won’t be new to the experts) but I hope it might be useful to somebody…

True, if the location of the document contained a forward slash. However, documents in the database, or in the filesystem, are not located in a path name reflecting the name of a tag in the database. In other words, none of the documents in my databases are actually in a location that contains a forward slash, even though all of my tags are organized by groups (with tagging disabled) that start with a forward slash.