Semi-structured documents

Are there any plans to add the possibility to add semi structured entries to DT? That is something I would love to see

Well, I can’t tell right now. Any examples?

Well, the general case is that I have some information that has a “regular structure”, for example notes on a paper. It would contain title, author and a reference (url link) to the reference in bib-manager. In a case like this I would like to define a “note type” that had 4 fields, title, author, reference and note, which I then could use… This would allow me to keep some structure of these notes (so I for example could extract the data later using applescript) instead of just one big note.

Other example would be to extend the address book, keep meeting notes (date, who were there, subject, location, notes), have project info, etc, etc.

I do realize that I could use an app like FileMaker or Bento for doing something like this but I think it would work better in DT.

What about using the sheets/forms of DT Pro? Their layout is of course very limited but they might be sufficient.

Yes, it’s part way there. The record part would work but I don’t necessarily want to have a collection of them (the sheet). For example I could see myself have a project group where I had a couple of meeting “records”, some paper records, a few people records, etc.

So except for the “being a part of a sheet”-thing records would work. I would like to see records being ordinary documents. (if I’m making sense to you)

(and it would of course be nice to be able to use styled text)

Ah! Some kind of stand-alone records… that’s of course a possibility but don’t expect it in the near future.

Will have devon 2.x the power to add fields to object? I still have to create a record for every document I scan and after I need to group the record with the scanned document. It is a dirty way, to do the thing.

Yes, V2 will handle meta data.

Oh, a first reason to buy the leopard beta :wink:


Support for semi-structured documents would be a fantastic addition and address a major issue I encounter with my work/knowledge organization and DT Pro.

Maybe it’s a relic of using card catalogs, but I basicly have carved up my knowledge world into subjects and themes (and as a geographer into regions). Essentially, I deal with information all the time that fits into multiple subjects, themes and regions. For example a paper about ICT in Europe may include information about ICT, computing, Europe, UK, Czech Republic and the EU. With the bibliographic software I used (and still must use) my workaround was keywords. Very limited, very inprecise, but the best think I could figure out.

With DTPro I’m sorting things out into folders, but the contents overlap. I can replicate and duplicate documents (I’d like to hire somebody) but even then when I search I’m getting every damn text with Czech Republic and ICT in it. Sure the score helps, but with hundreds of documents its turning into a muddle of documents. (The see also function is fantastic too in helping remind me of sources I’ve overlooked). However, every search now turns up so much, I’m getting overwhelmed trying to find the article I’m looking for.

What I’m thinking many of us need (and certainly I do) is a function to use AI capabilities just with keywords and structure our data in some manner analaguous with bibliographic software. Perhaps metadata will do the trick? I’m on the edge of my seat waiting! I really hope that DTPro 2.0 will support metadata profiles that can also be used to customize the interface for the presentation of items.

This said, and I could go on with wishes, I think pragmatically right now it would be good to find out how people go about connecting traditional bibliographic information to the documents they add into DT Pro to learn approaches to getting the most from the software.


I think every work that uses devonthink to keep documents that have a phisical place (like books, or paper documents) need a way to assign -to devon documents- fields to keep info about the author, date, place et ceterae, and a good tools to search into. Some days ago another office ask me about a program to scan and keep in order photos and documents and create queries of datas about the photo and documents scanned, and I can not say ‘use devon!’.

I would like to second the call for structured entries (fields, tags or whatever). The freeform structure of DTP works well in many cases (and often better than a structured SQL-like approach). But as others pointed out there are situations where structured entries are helpful (and why not use a structure if it is already there?). As I understood, metadata are on the way, so I am locking forward to it. I think combining DTP freeform AI with a structured approach will make a good evolution of an already fine tool …

In the UI I would like to see those structured elements not only in the info window, but (at least a subset of them, customizable at best) as (editable) column in various views (including search) and within the preview pane.


I would like to second that too.
I am new here. Just switched from PC to a 24" iMac, which I greatly enjoy. Devonthink was my main reason for switching.

I have been using “UltraRecall” ( as information manager on my PC, probably one of the best in the PC world. In Version 2 it already had custom metadata handling and upon user request (I was one of them) in Version 3 now they added semistructured entries. But it lacks in other areas so when I stumbled upon Devonthink, I decided to switch.

Now my biggest wish for DTP2 is a good custom metadata handling, which I have read is on its way. My second biggest wish would be, as Johannes suggested, customizeable structured fields within the preview/editor pane. In UltraRecall I think this was achieved by a combination of xml and rtf in one document. I can imagine (all I can do with my limited computer knowledge) this could also be done if there would be a word processor program (whose files could be rendered by DT) that can insert fields into the text that are linked to custom metadata fields. Don’t know if Mellel or any other program can do that, but maybe something like this can come to pass in future (hopefully soon!)


Ooh ooh ooh!

I’m not sure I understand this correctly, but I’m @#$%ing excited.

Would I be able to (through AppleScript, for instance) query a specific metadata field of a file?

set theValueIWant to metadata field "Author" of SomeArbitraryDocument

or even embed record data queries in an HTML/XML sheet “template”?

<query>metadata field "Author" of theCurrentDocument</query>

Even just with a simple layout there with AppleScript embedded in HTML templates, I think the possibilities are immense. DEVONthink would instantly become a simple but devastatingly sexxxxy FileMaker replacement.

Even if DEVON wanted to split off this functionality as a large plugin for DEVONthink Pro 2.X (and charge extra for it), I’d be excited and would cheerfully pay for the added development time.

Of course, if what you guys are implying involves much less (or different) functionality than what I’m talking about, please correct me.

As someone who has tons of analogue items in various forms and locations (eg: DVD in 'resources crate #4", articles from journal in file cabinet- file # 45) I too would love to see DTPO move toward offering something in this area - basically, the capacity to add fields (for me, every single item would need a unique, numerical identifier - plus a field where I specify the location. Add titles, keywords, author, date etc and it’s everything I need for info management.

Any hope of this or is it outside the scope of DTPO??

There is hope :slight_smile:

I once used a db called Paradox so I could have the benefit of only semi-structured info. It was the traditional columns with headers, which were one quick way to group the information for a search or for listing. Now I want to do a simple vocabulary list and cannot seem to find a way to do that simple kind of setup so that I can search by definition or by word entry. Paradox is RAM based and exceedingly fast, but a lot more money. Devon Think Pro seemed an intriguing way to funnel info automatically into some form of organization. I’m not sure that it’s all that forget about it automatic, but maybe I’m not yet with the program enough to see the best way to do things. All I know is that the ability to scoop a bit of text out of an html doc you’ve discovered by search of an RSS feed right into DTP is pretty cool. I don’t know about Paradox now …I’m assuming almost any db program worth its salt is or will soon be coming around to the model of “funneling data” to db files on some kind of set and forget basis, or, at least, in such a way that the ease of transporting data from the net to your db package will be accomplished without having to go in and out of various open applications and files. I imagine other programs will “borrow” conceptually from DTP. I also believe, however, the real freedom is to be able to establish db structure according to the info, which will vary from project to project. It just so happens that this may be beyond the conceptual basis for dtp. If so, I’d like to know how Devon sees itself. On the other hand, I’d also like to know if there is some way, other than with sheets, to take control of the structure, even override what the program seems to want to do. If I’ve missed the boat …and I will say, I’m still at that point where I’m trying to see how this program can work for me …kindly tell me that I don’t know what I’m talking about. But, please also tell me why! This is just my two cents.


Sorry, not Paradox, Panorama…Panorama …Panorama.

How long is the vocabulary list and how long are the definitions?

If it’s the sort of list you might use to learn a language – lots of words, brief definitions-- probably a spreadsheet is the way to do it. DTP has a “sheet” format.

OTOH, in my main database, I have a collection of things that are more like mini-encyclopedia entries: not very many words, but with detailed explanations including formulas and/or figures. For these, I’ve created a “terminology” group. Whenever I have to look up some semi-obscure technical fact, I create a document for it in this group.


I just wanted to include my interest in adding meta data to files. As with many of the previous folks, I have lots of things that need simple location tags and keywords. One of my issues with things like EndNote (sorry, but I still need the bibliography software for some things) I love using keywords. Clunky, but manageable. Unfortunately, one of my main problems is remembering the list of keywords that I’ve been using. I have a rather bad habit of re-thinking my keywords as I re-read documents, and at that point things start to fall apart.

The ability to develop scripts to help me search through meta data would seem a much better option (including a possibility for changing all tags from one thing to another - oh, the options!). It would also be nice to have a way of viewing just meta data. I think that would give me a better feel for my own documents and my current structuring system. Goodness knows meta data can’t possibly confuse my lack of structured documents any further!