I am one of those users who was asking for the graph view to show links between notes based on DEVONthink item links, wikilinks, tags etc. I know it’s not for everyone, but it’s for some of us who like to see the connections presented visually.
But I wasn’t a fan of Obsidian despite it having a pretty graph.
After previous comments in this forum, I wasn’t expecting DEVONtechnologies to ever implement it, so I was making do without it and doubling-down on tags as my core connecting device.
BUT, what a nice surprise, you have delivered. And it’s the reason I have upgraded.
AND, I’m loving it.
The See Also is a bit random sometimes, but is a nice addition. I was only expecting actual links to be available, but the See Also option provides suggested notes as well - nice touch.
So, thank you for implementing this and well done on the features!
I’m sure there will be requests to make it better (like the ability to drag points around to move them away from congested areas) but it’s a really great addition and now that the engine is in there, enhancing may be possible.
I’m also digging the graph, but the zoom speed is really slow. If I want to zoom in on the graph to read certain parts, I have to do the touchpad zoom gesture 8-10 times before it’s big enough.
I definitely meant it as feedback, not a complaint. I’m loving DT4 so far and know it’s beta still.
As for a real-world use case… nothing solidified yet. My plan/hope… I work in IT and deal heavily with vendor knowledge base articles. Much of these include code snippets or log samples of an issue. As my database of these grow, I’m hoping/curious to see relations and make DT my first stop for researching an issue before diving into Google.
In a nutshell, the real-world application for me seems to be that the graph provides an additional way of viewing connections that is easier for people like me who prefer a visual representation of the connections. But it’s also nice to have those list views as well, you wouldn’t want to just have the graph as it could get too congested.
1. Daily Notes
In my Daily Notes I use both Item Links and Tags to links documents together. I have some routines where every day I connect documents to my central Daily Note, such as all emails for that day, specific photos or documents. For these I use Item Links. When referring to certain topics in my note, I use tags. Yes all of these connections * can be seen in the Links Inspector or the Tag Inspector, but having the Graph view provides a nice visual map of the connections. My brain processes the map easier than lists.
I often like to link various PDFs to PDFs or to Markdown Files and other file types. I do this when they are related, which could be one document supports the other, or they relate to a certain project. The graph view shows these relationships when you might otherwise have forgotten the link exists.
I love that the graph view shows all link types - Item Links, Custom Metadata links, Tags and Wikilinks (which I don’t use but I know others do). That provides so much flexibility to support the different ways we all work and connect our information.
EDIT: I forgot “Mentions”. I don’t quite understand that yet, but look forward to exploring what it means.
3. Aiding research
The See Also is going to useful for research as it will make suggestions based on DEVONthink AI on possible links to explore.
I am looking forward to continuing to explore how else it will assist my personal and professional work.
I’m finding that tags are not showing in the graph for some documents, but ok for others. I cannot see a pattern yet. I have checked and can confirm that the tags do exist. This is occurring on markdown .md files, PDFs, and email .eml files.
Does the graph need to do some sort of background indexing before they show up, or is the graph generated on-the-fly when a document is selected?
Zooming is also possible via the keyboard shortcuts + and -. But I just noticed that keyboard navigation in case of a zoomed Graph inspector does not automatically scroll, the next beta will fix this.
The Graph is calculated on the fly but the number of nodes is limited and tags have the lowest priority. Maybe some of your documents are heavily linked and/or have a lot of See Also results? Screenshots would be useful.
For a real-world use, I’m finding it interesting in my research on medieval manuscripts. I have a database of transcriptions of/notes on hundreds of medieval manuscripts. The graph view is showing relationships that reflect direct links and shared libraries, but also unexpected and less immediately obvious relationships - via references to titles found in multiple manuscripts, or the names of individuals connected with the books in various ways.
In terms of what @TheFarSide already sketched, which is congruent with my use(-interest), I would propose consideration of following improvements/enhancements:
ability to detach the graph as floating window – reason: the power of the graph for me has always been dual: I - schematic visualition of relations and II - interface to meaningfully browse complex content repositories – of course, here I am thinking of the potential of ‘II’
give the graph a history (back – forward) – reason: see above; additional: it would really contribute to the coherence in navigating DT, which can become a challenge with multiple views, windows, panes, some window view states even being detached from current (main) selection etc etc. – the graph here could really grow into the role of a ‘navigational binder’ and make things coherently visible and cognitively relatable/traceable
allow toggle btw reference of graph to current db as well as to all open DBs, just as with ‘see also’ – reason: would deepen the usefulness of graph to a) discover content/relations in different perspectives; b) allow comparing informational patterns of different contextual situations
– the graph always was great, it has become greater, it could be a real ‘powerhouse’ for intelligently navigating ones system of DBs, and it seems everything is ‘almost already there’ (– though I am speaking here being ‘technologically blind’ as to the dev-challenges this really implies; but I think it would be well worth resources in this case!).
Not being the poster of that last question, I would personally read the “would it be possible to consider such addition…?” as the core part, or at least the more central part of that question.
So, while “future” is an inherently fluffy category in allmost all context – and often leading straight to opaque oracles talk – I think the underlying question is: are these stated FRs noted somewhere in the Devonsphere/customer-support system?
It would be nice to receive notice/acknowledgement on that, like in some form of simple clear acknowledgement.
This actually brings to the fore to a more general question here: is there some more technical protocol to be followed here in the forums for any such FR (or: proposition) to receive a “standing” as FR/proposition worthy of acknowledgement and ‘counting’? (– like opening a separate thread, certain standing formulas, or the like?!)
I arrive at this more general question – on the ocassion of this particular plea for clarification (acknowledgement) – because, by now, I often wondered what the precondition is ‘to be counted/considered’ as a FR is in the forums, or whether such exist?
As sometimes in threads I read things like “it´s noted” – while in other, structurally similar contexts (like here), I rather find more ominous answers or comments, leaving things open to ostensible ambiguity.
– What I am simultaneously & inherently saying/proposing, on top of this specific inquiry, is this:
It might be generally good/productive to educate the DT userbase/customers, as to how to formulate a real FR, worthy of ‘standing’, or more generally educate them as which channels to take in such a case.
I think this would contribute to avoid non-productive/non-sensical communications in the forum.
(I looked up the forum rules on this occassion, but only found more general guidelines like “Treat others like you would like be treated.”)
TIA for any clarification on the specific as well as on the general matter raised.
I just want to add to the comments about Loving the Graph. It is an extremely useful addition to DT. I use it to look for connections that could otherwise be missed. It would be nice to know what criteria are used to create the graph (what docs are chosen, etc.) and how it differs from See Also, which is far more broad and often chooses other documents as well, or instead of, those in Graph.
I am aware that See Also is a connection type. My question relates more to HOW the documents in the Graph are selected, as opposed to how the ones in See Also are selected. There must be some kind of algorithm that DT uses. Can you explain how that works, hopefully in simple terms?
