Is there any technique to create a link (not in the sense of a file system link of course) between two or more items in the database? By “link” I mean that I can tell DT that those items belong together, and that when viewing one I have simple and quick access to the other(s). A typical application would be to tie together book chapters or multipart articles.
Since someone is bound to suggest “use groups”, here is what I do not like about that solution: it interferes with my “group by topic” principle, which I guess is also important to keep DTs classification algorithm happy. Moreover, I could even imagine situations where the items to be tied together would live in different places in the DT hierarchy. An example would be a series of book chapters that discuss sufficiently different topics to be classified separately.
Another approach would be to create an RTF document with links to all the other items. Problem: I cannot find that RTF document back from any one of its items.
If I understand the original post right, this is something that could be a useful feature. What the user wanted was a multi-hierarchical way of grouping. Of course DT will have to keep the main hierarchy for display and auto-clasify etc, but wouldn’t it be nice to also be able to associate contents X, Y & Z in DIFFERENT groups together? In effect, this would mean any content could have more than one parent group, more than one group it belonged to.
In fact, I simulate something like this now for a few files using replicate, but it tiresome and multi-grouping would be better…
Actually, there is a way to do this. Here’s a scenario. I’m writing my autobiography. Try these tricks:
 Create a new group and call it My Life and Times. Within that group create a rich text document and name it Table of Contents. In this new document lay out the contents of your book, e.g., “Preface,” “Chapter One – Early Childhood.” Chapter Two – Teenage Years: College and Graduate School," “Chapter Three – How I Changed the World,” and so on. (People who write autobiographies tend to brag about themselves.)
 Make certain that DT Preferences > Editing > automatic Wiki linking/names and aliases is turned on.
 Now create new documents whose names correspond to the phrases in the Table of Contents document, e.g., “Preface,” “Chapter One – Early Childhood,” and so on. You can do this with Wiki linking to create the new documents from the Table of Contents text strings, if you wish. (Caveat: There must not be existing files with the same name already in your DT database.)
 Now you have the table of contents of the autobiography, and a series of files that correspond to the elements of the book. The text strings in the Table of Contents are hyperlinked to each element, e.g. the string “Preface” in the Table of Contents is a clickable hyperlink to the document named “Preface.” All that’s necessary now is to fill in each element document with deathless prose.
 But we have only one-way hyperlinks from the Table of Contents to each of the elements. So, at a convenient location or locations within each element, e.g. top and bottom, we type or paste in the string “Table of Contents.” This creates hyperlinking from each element document back to the Table of Contents. (Caveat: "Table of Contents must be a unique filename in your database. If not, change its name to make it unique, e.g., “My Life - Contents.”) Now we’ve got two-way linking between a “master” document and its elements.
 It doesn’t make any difference where you group the element documents. Suppose, for example, you want to move “Chapter Fourteen – How I Proved Einstein Wrong, and Revolutionized Physics” into your Physics > Special Relativity group, and still another chapter into your Molecular Biology > Prions group. No problem. The links will still work to tie your autobiography together!
Thanks for the detailed description. That method should work fine for authors. However, the scenario I had in mind is different in one crucial aspect: I was thinking of collecting data from elsewhere, i.e. PDF files, which I cannot edit, so I cannot add the backlinks to the table of contents. Or is it possible to put links into the comments?
Technically, using replicants to store contents in multiple groups is identical to “multi-hierarchical grouping” (both contents and groups can have multiple parent groups). Therefore the real question is how replicating could be improved/simplified (as linking is not bidirectional and limited to rich text documents and therefore no complete solution).
From my point of view (keeping in mind my original scenario), this is not quite the same, because the database hierarchy also serves for classification. What I was looking for is a way to group documents together without any impact on classification.
If replicants were improved or simplified, or multi-grouping introduced, that might have very interesting effects on how people use DT. As far as I know, there is no current program out there that can (easily) organise data in other than the usual one-parent-per-content structure - the usual hierarchal model, in fact. Now, perhaps the reality that people’s DT database represents IS hierarchical, in which case, no problem… but on the other hand perhaps they are forcing their data into a hierarchical model which isn’t really natural for it, just because they have to. To an extent it is obvious that there’s a need for multi-grouping - that’s why Devon have provided replicants and labels (another way of attempting the same thing). The question is, how big is the need?
If Devon did introduce this, it would be… interesting.
If you’re using lots of replicants, then the structure of the database isn’t hierarchical anymore - you’re creating a net. Therefore the only issue might be that there are no other views (e.g. maps) available yet. But they’re in the pipeline
Yes, a net is a better description in principle, though in practice it’s a tree with occasional cross-links (i.e. it is almost hierarchical).
My “dream” structure would be a general graph in which which each document is a node. Groups would then simply be empty documents. Yes, I do realize that this is probably not so great for classification…
Classification shouldn’t be an issue as the mentioned structure is somewhat similar to the internal structure. But probably a lot of people would be confused by such a structure as they’re used to files & folders whereas documents with childs or groups with contents are (at least currently without additional views) not that useful.
Not sure if this is what you’re asking for, but my old Hypercard stack of research notes had a space on each card for various categories that I would assign by keyword (place, date, musical style, friends, etc). then I could have the whole stack organized by whatever keyword I typed in. So a card that applied to choral music in San Francisco in 1938 could be viewed in a group with everything else that happened in 1938, or in another group with every other card related to choral music, or in another group with all the cards related to San Francisco. This really helped a lot in keeping continuity among various threads.
I’d love it if such user assigned keywords could be used in DN, but there’s probably a way to do this or something similar already and I just haven’t worked with DN enough on my big project to have figured it out yet, or even if I really need it anymore, given the various links possible in DN.
I suppose the same thing could be done by a regular search in DN, or by replicants, and I’m also thinking of assigning a label to the major categories so that I can then sort by label. But I haven’t tried this yet so don’t know how it’ll work, and in any case there’s a limit to the number of labels.
The application Boswell might allow this, too, but I’ve not used it.
You can just type your keywords into the comment field of each document. But the keyword system doesn’t scale well to large databases, because there is only one “level” of keywords. You also must be careful not to mistype a keyword, otherwise the document just seems to disappear from its keyword category.
I definitely would appreciate the possibility to create back-links in some way.
Wiki-Links: instead of just creating a blank page it should add a backlink to the place it was created from.
It would also be cool if on could create “hard links” between documents just as mentioned before - and thus providing DT with some other form of grouping/relatedness. I mean it could be possible that DT would never see a relationship between the files but I (having extra knowledge and a reason) decide to link them… maybe with a keyword that, if encountered elsewhere would associate those items as well?!
Oh and re: the “net” feature - I cannot wait to see it!
Whoa–wide avatar, Dekay.
When I have two closely related rich-text documents (for instance, one might be a rebuttal of the other), I will insert text or a phrase (“Relateded Text”), select it, control-click and choose Link To. There you will find a pop-up of all documents in your database.
Would this satisfy your desire for a “hard Link”?