replicate a tag -> replicates not shown in info window?!

I have tried re-structuring tags in a database:
I sorted them into groups and some tags belong to more than one group.

I’d like to see, where tags are placed in the hierarchical tree structure, but the info window of a tag does never show any replicants, even if I have replicated the tag to several tag groups.

Is this a bug or a feature?
[size=50]I’d vote for the 2nd option.[/size]

No bug. Tag groups behave differently than “normal” groups.

Just curious, but are you back to doing all this group nesting that you were doing when your recent cascading chain of unfortunate events began? :open_mouth:

yes, I still try to get structure and overview in my database (and that requires sorting the tags in hierarchical groups).

I have to work with my database and have to find the most relevant documents for my thesis.
And that requires a lot of sorting etc.
I wanted the documents to be tagged in a way that I can find with the corresponding tag all information (created by myself or others) concerning a certain sub-sub-topic of my field…

… I think the reason for the unfortunate events was my lack of knowledge about the details of tagging etc., but there are still some things in the “behaviour” of DTPro which are quite confusing to me.

I’m not an expert on tagging, nor do I expect that anyone is as there is no ‘right’ or ‘wrong’ way to tag documents. If I had more time this morning, I would go into greater detail, but let me say that some tagging strategies do work better than others. My approach to tagging is that the tag structure is flat, while a folder structure is usually hierarchical. A folder hierarchy allows you to file and retrieve documents when you know where they are, or where they should be. Tagging allows you to search for documents that are related in ways you may have forgotten about.

Tags and folders each have their strengths and weaknesses as a tool for organizing data. Most In my experience, when you try and blend tags into a hierarchical structure, you get the weaknesses of both. A deep tag hierarchy becomes either not useful or too time consuming to maintain, or both. Please understand that I am not saying that one cannot, or should not, use both a folder hierarchy and tags-they actually work well. I’m specifically suggesting that a deep hierarchy of tags is perhaps not the best approach.

I’d suggest that you first try tagging in a flat structure and get some time actually using DEVONthink instead of constantly tweaking it. This will allow you to become more familiar with how DEVONthink can work for you, and also allow you to learn more about your data and its relationships. Then in the future you can tweak things to better work for you, while understanding the dynamics of the application so that you know with certainty what the results will be when you do make changes to the database structure. I’d also recommend that you get a copy of Kissel’s book on DEVONthink, available from the DEVON Web site. It’s an excellent primer on getting started with DEVONthink, and for the $10US investment, it will pay you back in time saved.

1 Like

Here is an article by Brett Terpstra that may be more beneficial anyway. I don’t follow the exact tagging strategy that Brett does, but I’m close. Look in particular at the mind map at the end of the article.

Hi Greg,

thanks for your answer.

I see that the nested tags structure has its weaknesses.
However, I started with a flat tags structure and got totally overwhelmed by it, because the tags were sorted alphabetically and I had no overview at all.

So for me the solution seemed to be a hierarchy for tags.
Maybe DevonThink is not planned to be used that way?!

My database contains thousands of notes about scientific literature about manufacturing (especially the process I’m working with) and excerpts and comments from me about some of them.
And as for me it is important which of the many process parameters have what influence on which part property, tagging got quite complicated and I finally decided to give a tag for each process parameter and each part property, each material and so on…

So I hope I’ll be able to quickly display e. g. all sources which are about a certain parameter or a certain product property.

I now had to sort them in a way that I have the right documents at my fingertips in the right moment.

Unfortunately I’m under high pressure and I do not have time to try different systems and see what fits best.

Many of my problems came from the fact that I was not aware of many of the subtleties when moving items, removing or renaming tags etc…
So I’ll have to try if it works…

Thanks for the article about tagging and for the hint on the book.
I’ll gladly invest the 10$.
It maybe comes some days too late, but I’ll see how to make the best of it.

I believe that DEVONthink has a lot of power and flexibility to be used just about any way the user desires. With that comes the possibility to complicate that which would be of more utility if left simple. On one end of the spectrum, one can create a thousand different tags and apply only 1-2 to each document, or at the other end of the spectrum create fewer tags and apply multiple tags to each document. I prefer the latter, as I don’t have to remember a thousand tag names to search on (see Brett’s article for natural tag names). If I don’t have a thousand tags, then I also don’t have need to organize them in a hierarchy.

Greg, I see what you mean.
But I don’t see how I should solve my problem with less tags.
(Sure I could rely on smart folders with complex search criteria, but as the documents are in at least 3 different languages, this would be very tricky.)

I have a manufacturing process with 4 major processing steps which have (each one) about 5…15 important input variables (pressures, temperatures, velocities, heating/cooling rates, material properties, etc.) which can affect about 10…20 basic properties of the produced products in the 3 intermediate or final stages of the product…

I don’t see how to do that easily.
And so I’ve nested the input variables into each processing step they belong to, etc.

(maybe I am complicated too :frowning: but it is indeed a complex problem I’m dealing with).

This sounds like a situation where a combination of groups and tags might work better.

I’m guessing that individual references don’t neatly fall into a single category. You might have a paper that talks about the heating/cooling rates and the material properties, or about one particular sector of the process space. So you can’t really group the “pressure” or “temperature” papers together. On the other hand, as you’ve discovered, a flat tag structure doesn’t work either.

So what about creating a group for each of the major processing steps, and perhaps another for each of the stages of the product? Then use tags to sort out which papers deal with which input/output parameters. Possibly include enough information in the tag to say which process it belongs to: “Process A–temperature” may or may not have much to do with “Process B – temperature.”

Your own notes can also serve as indexes: you read four (or forty!) different papers about Process A, and from that you tease out an analysis of the relationship between cooling rate and crystal structure. So you write that analysis out and include links to all the papers you referenced. (And it might make sense to tag it as “thesis,” since it could easily serve as the basis for a section.) If you’re really organized, you could even use the Wikilinks function to tie several such analyses together.

As the database gets bigger and more unwieldy, you’re also likely to find the See Also function helpful. It’s not as comprehensive as the search, but is a good way to make sure things don’t slip through your fingers just because you didn’t get the search terms exactly right.

Hope this helps,


Hi Katherine,

thank you very much for your long comment!

You’re right with what you say.

The problem from my point of view is:
The 4 major processing steps are highly interconnected and process parameters in step 1 can also show effects in steps 3 and 4.

So, most literature also touches more than one processing step - however (that’s right) some focus on one of the 4 steps…

In other words:
when “filing” the papers in the folder structure e. g. for the different processing steps, any documents would be replicated to different folders (but that’s possible and that would be ok).

I did what you proposed, but only with tags up to now.
I’m not sure if it really makes a difference if you use only tags or tags+group structure (besides that it makes the tags folder much bigger…).
And for sure the names of the input variable tags contain not only the physical symbol/name (p for pressure or T for temperature or t for time), but also an abbreviation for the processing step etc.
(In step 2 there are at least 2 temperatures, 2 pressures etc…)

I think after solving the “technical” issues with DevonThink (and many of them even were not due to DT but to the existence of hundreds of duplicate files which made things even more miserable), the tags structure is more or less ok and can evolve while I’m continuing to work with it.

The big problem is writing (as you address it correctly).
I don’t have a clear system yet, how to make notes on sources and so on.
I have a big BibTeX database (>2700 records) which keeps the bibliographic data for many (but by far not for all!) the sources.
And those sources, which are already in the bibtex db are consequently named DeVille2011b.pdf .

However, there are many sources which do not yet exist in the bibtex database and therefore do not yet have a “unique identifier”.

I’m not sure what’s the best way for taking notes.
For highlighting interesting passages in the pdfs, I’d like to use Skim.
However that means, that I can not edit the pdfs directly in DTPro but have to open them externally. (That’s ok, just another extra step.)

Furthermore I made a lot of RTF notes about contents of papers or with copied sentences or paragraphs from a document.
Their name starts with the identifier of the Source: “DeVille2010c: …”, so I can easily see where they came from, however ”see also" ignores the information in the file name AFAIK (which might be a drawback and it might be better to include this identifier in the document itself or anywhere else…)

I’m writing my thesis with LaTeX so I’d prefer to type in the text editor directly and not in DT Pro.

… it’s difficult to establish a workflow which makes sense.

And as I already emphasized, time is very short and I definitely must be very productive from now on. :frowning:

So if anybody has good ideas or examples how you do your (academic) writing (ideally in combination with LaTeX, BibTeX), I’d be glad to know how you do it.

Today I bought the ebook about DTPro which Greg_Jones recommended and I’ll have a look if it helps me…

In the last weeks I experimented a lot with different approaches, but I’m not very happy with it yet:
I found Freeplane (an excellent Open Source Mindmapping software) a great tool and its brother SciPlore Mindmapping also adds features for referring to pdfs and bibtex references. (And Freeplane can even add mathematical formulae with LaTeX code to a node!)
But what seemed great at the beginning was not as “smooth” as I wished it would be and I could not find the Mindmap well suited for really writing long paragraphs and text.

I’d like to know how other people are writing their thesis.
My idea would be (for the “state of the art” chapter):

  • reading the important sources.
  • extracting the important points from those sources
  • re-arranging them/their order to fit in the structure of the thesis (important: still knowing which idea came from which source)
  • writing the text “around” those points.

For that reason I’d prefer a structure where each single “atomistic” idea/point/information is in a separate small document (which still points to the source).
The branches of a mindmap are not bad for something like that but I could not really get into liking that…

Kind regards


p.s. “see also” is very useful (and gets more useful the more the database grows).

What I found I’m missing in my workflow with DT Pro is also the following:
to mark an idea or snippet of text as “already used” in the thesis…
I’d like to have a way to display only those ideas which were not yet used in the big document, but I did not find a good way for that yet…

I’m not sure the difference between tags and groups matters to DTP, which treats them as pretty much the same. But thinking of them differently can sometimes be helpful to the humans.

For links between notes and references, I’d include a link in the body of the note and/or in the DTP metadata. That way you’ll still be able to find the reference when you’ve forgotten what DeVille2011b was, and you won’t have to be as concerned with identifiers. I’d also suggest pulling excerpts and such out to separate files. You’ll thank yourself later, as that will make them easier to find.

You can flag bits as “used” using either the Flag field or the Label field in DTP. Either of these can then be used in searches.

For the writing piece, go download Scrivener. It will change your life. ( (I haven’t personally done Scrivener -> Latex exports, but I know there are Scrivener users who do, and information in their forums on how to do it. )


Use labels. You can change the description of each label to fit your needs in the preferences. Plus they’re searchable.


I know about Scrivener and the concept looks great.
Several times I thought about trying it, but I did not, because:

  1. I think it is already too late to change the Texteditor. I have only some weeks left and all the configuration and learning is too high a risk and will take too much time IMHO.

  2. I use a lot of special commands and “non-standard”-stuff in LaTeX and I don’t know how good that would work with Scrivener.
    For my next phd thesis (beware!) I would maybe try such a tool.
    But this time, I fear, it’s too late.

I admit, I never tried it (and everything looks extremely promising, e. g. the cork board and the possibility to add a status “draft”, … to sections) and I’ll think over it again, but I think I have already too many different tools to handle - one more would make things more complicated

Thanks, interesting idea with the labels.
The problem is:
My notes are quite lengthy (and not yet split in “atomar” informational chunks), so maybe I’ve used only 1 thought of many in a rtf.
Then I can not mark it used…, ok I could have a “partly used” and a “used” label… hm. I’ll consider the labels approach.
In the RTFs I was trying to highlight the parts already used with a certain color, but it was too difficult to make that work in DevonThink (too much mouse clicking )

Interesting idea. However will I be better with more than one database?
I’m not sure, because I think that will enlarge the “chaos” and reduce my overview…

Thanks, Greg, for the article.
it’s an interesting read and the emphasis on “simplicity” of the tags is a good point…

p.s. (short update):

thanks again for the proposed approach with the labels.
Somehow I thought that the labels in DTPro are identical with those in the finder (and therefore did not want to use them for such things).
Now I found that they seem to be independent of the finder labels and will try to mark the documents as “to use”, “partly used” and “completely used” with the labels.

Furthermore I found the freeware Notational Velocity which can edit text or RTF files and can also deal with DevonThinks OpenMeta tags (import/export).
So maybe for some kind of notetaking this could make sense, as NV can create one text file for each note and I then can index the notes folder with DTPro and they are also available in my database.

concerning Scrivener:
its features are very very tempting, but I’ve decided that it is absolutely too late in this project to try it and/or change my workflow and tools that drastically.
And that would be one place more to store notes and ideas.
I’ll try to stay with DTPro for the contents and my LaTeX editor (TexmakerX) for the editing/layout.

just a little update:

I’ve changed my workflow for annotating and tagging information from sources in pdf files:

long explanation in the other post above.
short summary:

  • using Skim for annontating and highlighting pdfs.
  • I created labels “not used”, “partially used” and “100% used” to mark items, whose information already was used in the thesis.
    The disadvantage: those labels will not make sense for whole pdfs, as they nearly never will be 100% used, so I found the following solution:

if information in a pdf I’m reading in Skim will be relevant for a certain chapter or section in my thesis, I will no longer add a tag “use in chapter x.y” to the whole pdf, but create a green highlighted note in Skim and at the same time a rtf note in DTPro, which I can tag with “use in Chapter x.y” and which I then can mark as “used” by a label, when its information was used in my thesis.

(for more information see: viewtopic.php?f=20&t=10929&p=59725#p59725)