I prefer to always keep full source information (in APA format) in the same document as any quotation, so that, even if I paste the text to another app, I never risk losing track of where I obtained the information. With short quotes, that often means that more than half of the text of a document is source information. Jeff’s post has got me a bit worried. Should I amend my practices, and is there a good alternative to what I’m doing now, preferably without relying on links to other applications?
Background: Most of my notes are generated from within Sente using the Dana Leighton script (based on a Robin Trew script) to export notes to RTF/OPML. That script hasn’t been updated in several years, and Sente has also been abandoned, so I’ll have to eventually migrate to a different system, but for now it’s all working fine.
It seems like this should be your answer. If your current process is creating an issue, then examine it. If you’re not running into problems with how you’re operating the software, I’d say you’re fine.
I shouldn’t have included that background information because it detracted from my main point. I don’t really know how useful/relevant I should expect See Related Text to be, so it’s difficult to know if it’s working as well as it could be. I’ve never felt shocked and disappointed that all the related documents thrown up were totally irrelevant, but on the other hand it’s rare to find anything that surprises me in a “Wow, I’d completely forgotten I had that, but it’s just what I needed” way; I could have found the same items by browsing through the folder containing the original item and a couple of adjacent folders.
No way to know for sure until I try it, but if, for example, I moved the source info (which is the same for every document within a folder) into Spotlight Comments, for example, am I likely to get improved results? I’m hoping for some indication as to whether Jeff’s characterization of the See Related Textalgorithm and implications thereof is accurate. For example, if it has some clever way of distinguishing between propositional content and things like APA references that it filters out, I will continue doing things in the same way. If, on the other hand, I’ve been inadvertently sabotaging the functionality I’ll change my practices here on in, and gradually redo the items that are already in the database. (My database is still not large, so the cost of switching is quite small. I’d appreciate an answer that would double as advice for someone starting a new database of this type.)
My approach to this problem is to use a temporary citation after the quote. This is likely to be in the form {Smith 2018 #1234}. It depends on which Bibliographic database you use (Bookends in my case). This has the advantage that if you paste all the text into something you are writing, it will be ready for the Bibliographic software to create the references and bibliography for you when you have finished.
This is not exactly a link to another program – it is a way of having a Unique ID number for each source that will permanently identify it. You can, of course, create a genuine UUID for the source, and use that.
I imagine that if you have a fairly small database, much of the material will be related, and you will not make any surprising discoveries until it gets bigger and the material gets more diverse. DEVONthink becomes really useful when you have so much stuff that you can’t find it by browsing a few related folders.
I’m not sure Jeff was referring to “See Related Text” or if he was referring to “See Also” – 'See Also" takes the whole document into account when searching for related material, and “See Related Text” does what it says – finds documents that are related to the currently selected text. It goes without saying that if the selected text was “he said” then you might get a lot of off-topic suggestions. If the selected text was “In Liege soon after 1945” then you might get spot-on suggestions. Try both See Also and See Related Text.
I do not believe the AI takes Spotlight Comments into account.
Which is also why @mbbntu’s comment is relevant. You might need to grow the base before See Also/See Related Text become more robust for you.