Another thing to bear in mind with reference managers is that, as far as I know, none of them are made by humanities majors. That means there tends to be better access to things like PubMed than to sites like Cambridge Journals Online, Oxford Journals, or Sage Journals. So far, the best combination I’ve found to minimize manual data entry is CiteULike.org and Bookends: CiteULike has a very large library of plugins that allow you to add references to your library by clicking a bookmarklet; and the Bookends Browser can navigate to any CiteULike page and detect and import selected references.
I’m in the same situation as the TS: looking for the right workflow including a reference manager.
Although stability is important, I feel very tempted to go with Sente + DevonThink Pro + Scrivener. However, I’m still wondering how this workflow operates in practice, especially in an academic setting:
-
Do you import whole articles in DevonThink Pro or do you do as Steven Berlin Johnson advises: enter chunks of text (max. 500 words) and selfstanding notes, ideas, hypotheses,… This would mean you copy the interesting parts from the articles you read, like he does, but don’t import the PDF’s of the articles (you would then probably use your reference manager for that)
-
Even if you can import your Sente notes in DTP, they are linked to certain articles, paragraphs in Sente, but they lose this context in DTP, don’t they? Isn’t this confusing?
For over a week I’ve been trying to figure out a cohesive workflow, but it seems that the only logical way would be to use one application for both gathering information and references… Can someone explain the philosophy behind having seperate apps?
If your do go with Sente, see @houthakker’s posts on the script he has written to export “notes” from Sente into DTP. It is a brilliant script that works wonderfully with Quicksilver triggers thrown in (you will pretty much get the SBJ effect, if that is what you are looking for).
DTP and Sente work well together and with the upcoming Sente 6 promising near realtime syncing across computers, things will be more interesting.
I have been using both Scrivener and Mellel trying to find the best system. Scrivener is great for working on large structured texts (theses, dissertations, books) but then must be exported to in another format in order to use Sente’s bibliographic scan. I don’t find this optimal at all. And while I love Sente for what it does, I would happily switch to an DEVONTech citation/reference manager app, as I think many users would.
Thanks for your reply.
So you don’t use DevonThink to manage all your PDF’s, but only your notes? Steven Berlin Johnson seemed to suggest that the AI system wouldn’t work that good if you entered whole articles (rather than quotes, notes, ideas).
As for the cooperation Sente - Scrivener: I hadn’t thought about that hickup for the bibliography; I was mainly thinking about citing and for that you can just drag the reference from Sente to Scrivener.
Sente’s reliability is also quite an issue – already had 2 crashes since I downloaded the trial version 2 days ago.
Is there any chance for a DevonCite to become reality in the future? 
Comment about Johnson’s technique of working with short snippets in a DEVONthink database:
I’m sure his technique works well for him, and there are undoubtably some advantages for his workflows – which he has noted as having assistants extract snippets from books and articles. 
But I don’t bother to split the tens of thousands of references in my databases, nor do I find it a necessary approach for my purposes of finding and using information.
I’ve got a number of book-length documents, some text, some PDF. I try to get familiar with them; indeed, I’ve probably put most of them in my databases because I’m already familiar with them. So when one pops up in a See Also list I’m not surprised, but may take a look at the Table of Contents.
Flip side of the coin: Suppose I’m really interested in a chapter about a specific topic, in a book. Do I invoke See Also to see other material that may be related to that topic? No, because See Also would try to decide what other materials are like the entire book. Instead, I’ll select a block of text and invoke the contextual menu option, See Selected Text, as that’s the subset of information in that book that I’d like to explore in other documents in my database.
Bill,
Illuminating reply, thanks. To put things in a context: I’m starting a PhD in law within two weeks. I’ll be reading lots and lots of case law and articles over the next 6 years. I was wondering whether DevonThink would work better if I only entered memorable quotes from those texts or if I could just drag them all in there – perhaps sometimes even without having read all of them.
I would also like to thank you for the notetaking-in-an-autonomous-rtf idea, just great!
I am currently using DTP to sychronize index the Sente folder after an hour or two of work-so, essentially, I let it manage the pdf library but only insofar as they are already done so by Sente. This might just be causing unnecessary database bloating, but it allows me to work within the Devon environment with the notes script from @houthakker. The downside of this might be that later versions of either of these apps become incompatible–I don’t know–but for the time being, it produces something like continuity between both.
On the question of AI and semantic or conceptual linkage – I am still skeptical on the execution and rarely use this to mine conceptual cross-references from texts (probably just some lingering humanist fallacy 
).
I’ve done that, but really at this point in my work, less (in my workflow) is more.
I try to mention it often.
I think I’ve mentioned this before, Bill, but the meaning of See Selected Text is not at all clear. After all, I wouldn’t be able to select the text if I couldn’t see it.
A tiny point, you may think, but I think it’s an important one. The “Selected” bit can be taken as read (I wouldn’t have selected the text if what I’m trying to do wasn’t related to it), so how about “See Related Text”?
That seems to be a useful name, I’ll change this. Thanks.
Sorry for the previous post, which I hope won’t cause the discussion to veer off course.
It seems to me that it would be unfortunate not to store all your PDFs in DevonThink: it allows you to do full-text searches and gives you an out if you’ve neglected to take notes on something that later proves to be very important.
However, there’s no rule that the full-text PDFs have to be in the same database as the more selective snippets and notes. I’m leaning now towards going for the best of both worlds, one database for the lean-and-mean unadulterated Johnson experience and another for the long, heavy full-text PDFs, populated by indexing my Bookends/Sente attachments folder.
I’m not sure if it’s been mentioned here before, but it’s not difficult to get something similar-ish to the houtthaker experience with Bookends: Select a notecard, press Command-Option-Shift-L (it might be worth changing that shortcut), press Command-Tab to get to DT, create a new note, then Command-V to paste.* You’ll get the text of the note followed by a link that will launch Bookends (if necessary) and take you back to the reference. Of course, it doesn’t really compare to the smoothness you get with the Sente-to-DT script, but it isn’t as time-consuming as this description suggests.
*An alternative is to Command-Option-Drag a notecard to a DT note.
And now there’s the possibility to exlude huge PDF documents from Classify/See Also of course too.
Like these memorable courtroom quotes where the Q is a lawyer and the A is a witness:
I’d always assumed that that would also exclude the contents of those documents from being found by See Selected/Related Text. Is this correct? If so, the motivation for putting the documents in DT at all seems rather weaker, which is why I’m leaning towards the 2-database solution. Is my reasoning faulty here, or are there considerations I should be aware of?
That’s correct.
Sente to DT through Skim.
Hi all. I use another workaround for the Sente to DT problem, but I unfortunately dont know how to applescript to make it automated. I am doing all of the below manually now, and was wondering if someone could help me out with two really simple skim scripts!? Also, this is all based on the fact that I prefer Skim over Sente to read and annotate pdfs.
A. Script in Skim that says: if Skim opened from Sente (or even on every opening), create new text note/anchored note.
B. Then utilize the Services menu as Services->Sente->Insert Citation (Annotated Bibliography) into the note. Since you are opening from Sente anyway, Sente should be on the same reference, as long as it runs the script on opening. You can actually have the citation include all your Sente Notes, Keywords, whatever you want, by setting up a new Annotated Bibliography format in the bibliography manager in Sente.
- A second script in Skim that says: When saving Skim, automatically save .rtfd backup (Skim already has option to save skim notes, but these look bad in DT).
And that’s it! DT indexes the Sente folders automatically, placing the new rtfd file with the complete Sente reference and metadata (according to the Annotated Bibliography output settings in Sente) including all your notes and keywords from Sente (if you add them into the Annotated Bibliography), plus any highlighted text, notes, boxed passages, annotations etc. from Skim into one simple .rtfd file that is stored in the same local folder as the original pdf article.
Ultimately, basic users such as me can thoughtlessly download refs in Sente with all the associated metadata, open up the file in Skim, annotate and save the file like normal, and later use DT to search the pdf, the metadata created in Sente, and the notes/highlights from Sente and Skim. DT should also have the pdf and its associated .rtfd notes (from both Sente and Skim in one file) in the same folder.
Basically, I would love it if I could integrate Sente to DT with those two simple scripts that I am now doing by hand for each reference.
Thanks for any advice/thoughts!
Louis
With large documents (e.g., book sized PDFs) you get the possibility of matching quite a lot of searches. This has both good and bad points. Of course, as mentioned, you can just exclude those from the search/classification.
My general thought is to leave everything in and see the possible matches. You never know when something might be relevant. For really long documents, another possibility is to break it up into multiple PDFs (e.g., on a per-chapter basis) and group them into a folder. That way, your search/classification gets narrowed down to more relevant documents.
Another possibility is to mix the two approaches. Drag all the documents into the DT database but exclude the big ones. Put each of the big documents into a group with the notes on each document. I’d break things up into notes that are quotes in RTF file and personal commentary/thoughts in a second RTF file. That way, a search can tell you if it is a hit on the document itself (via the quotes RTF) vs. a hit on your commentary/thoughts. If you want to get fancy, you put each of the relevant portions in a separate file so you can use WikiLinks to link from the commentary to each portion directly. I don’t think you can link to a document section inside another document via WikiLinks.
I actually use BibDesk to store my references and link to the files inside the DT database. I haven’t found a convincing reason to switch to another package.
I’m very interested to try @houthakker’s script that sends Sente 5 notes to DEVONthink 2.
Sente 5 notes to DEVONthink 2 script
http://web.mac.com/robinfrancistrew/Site/Sente_Notes_2_DEVONthink.html
However, I Sente 6 uses a .sente6lib as the file extension and Sente 5 uses .Sente501 as appears below in the script.
property strSenteDbSuffix : ".Sente501"
I tried substituting .sente6lib for .Sente501 but it did not work. Perhaps the Sente 6 database structure has changed. Has anyone tried this? Would it be better to pose this question in the Sente support forums?
I’m just getting started with DEVONthink and am excited about using it for dissertation research and writing. Thanks for your help.
Will
I have updated my Sente to DEVONthink script
- To work with Sente 6
- To create bookmarks in DT2 to Sente attachments (PDFs etc)
As before, it exports Sente reading notes for a particular reference and page number(s) to RTF files in a folder named after the reference. (The folder is created in the DT2 global inbox).
(Bookmarks to attachments are created in the same folder)
The script is at:
Hi,
I was very excited to try the script with Sente 6, but I just can’t get it to work. I am new to this and am sure I must just be doing something wrong, but when I run the script I keep getting a message window telling me that I need to install Late Night Software’s XML Tools osax in a folder with the path ~/Library/ScriptingAdditions. I have tried doing that at least 10 times now, but to no avail. When I click OK to make the message window go away I then get another message saying “Nothing selected in Sente.” Can anyone tell me what I am doing wrong??
Thanks very much!
This thread fell off my radar for a long time - sorry that I didn’t see the November post, and I hope that issue was resolved. (I do get alerts from the relevant thread in the Sente support forum) https://sente.tenderapp.com/discussions/suggestions/34-export-notes-function
The script has evolved a little since the last posting here. It now avoids creating duplicates in the target folder, imports any PDF annotation snapshot graphics in the quotations field of Sente notes, and contains an editable switch, which, if set to true, will limit the export of tags to user-assigned tags.
The script is still at http://bit.ly/Sen62Dev