I wish the backlink pages was completed with the paragraphs of the original source
I am a postdoc researcher and I am also trying to collect my literature notes in obsidian. As some have described before, I create a new md file for each article.
However, I’m not sure if my linking and the use of my # is really useful.
I currently link my read articles with (Author Year), as I will cite them later in my paper.
Every article appears on a certain map of content (MOC). The Index contains all MOCs. I hope this makes somehow sense to you.
Attached you find my heading for a certain article. I am happy about every suggestion for improving my system
Why not link: both authors alone, the journal, and the year.
Basically the point being that you could in the future, filter by year, journal, or authors as well. I suppose you could just do this with Regex, but isn’t the benefit of bi-directional links that you could pull of the journal title for instance and peruse the backlinks to identify all the articles that come form that journal, or author, or year, as the case may be.
Just some thoughts.
Thanks for your input.
I thought the author-year combination is all I need to keep the system as easy as possible. But I see your point and I think you are right that it might be interesting to have more information, especially after a certain time.
Does it make sense to keep my author-year links and only add the other links within the article link (see picture above)? Or should I use the single links continuously and do without the author-year links?
If I look at it correctly, then I create some empty links, right?
Thanks again for your help.
I have not been publishing research for a few years, but I am returning to it. So, I am updating my research and expanding my PKM process.
Here are the apps that I have been using (partially updated) and their roles. I usually use a Mac.
Source searching: Papers 3.4, Google, Google scholar, etc.
PDF Converter, OCR: Searching for something new
Citation Manager / PDF storage: Papers 3.4 (Legacy) exploring options.
Knowledge and Note management: Obsidian, may also use Devonlink 3
Content outlining and drafts: Obsidian, and possibly Typora
Final reviews and publication: Word or Scrivener depending on size
Preferences: I prefer to avoid Platform as a Service (PaaS) Internet apps and subscriptions. The reasons for these preferences are topics for another discussion. But, while I am not uncompromising, I’d like to keep these two types of apps to a minimum, which is one reason I like Obsidian.
Searching for source materials does not seem to be a problem. I often have more than I can use.
For pdf conversion and OCR, in the past I have used Adobe Acrobat, but they went to a PaaS subscription basis and de-supported and killed my stand-alone app. Everyone else seems to use the ABBYY FineReader engine, so they are all about the same. The problem I have faced in the past is pdfs are not good at maintaining structure, so during the highlighting annotation process, accurate highlighting and quotes can be problematic and distracting. I’d like and OCR system that allows me to easily correct it. Someone suggested converting to HTML? I am certainly open to options here.
Citation manager: Mekentosj Papers 3 before they were bought out by Digital Science & Research Solutions (Readcube) had its issues, but they were constantly improving and addressing customer complaints, and it does some things very well. However, as a legacy app, I hate to rely on it for much longer.
I have tried Bookends. it is OK. I have used Mendeley, EZBib, ReadCube, and Zotero, besides being PaaS apps, they were awkward to use, but I will review updated versions of those. Anything else?
Annotation I heartily recommend Highlights app for use with Obsidian, https://highlightsapp.net. It is just so easy to use. I “Open with…” Highlights from Papers. Highlights, underlines, photos, and tables go into a sidebar with metadata and a back link to the pdf in Highlights. A key feature for me is that I can add a notes to each item I annotate. These notes become the text for a content outline and are essential to learning. Highlights has a handy feature for linking citations from each article’s references section. I can add tags and links for Obsidian here, if I want. I paste links and citekeys from Papers 3. Then, it exports to markdown. Obsidian reads it beautifully. I add a any more links and my literature notes are done.
Adding links and backlinks has always been a PITA, so I am planning to have Obsidian to a lot a heavy lifting in the note management, content outlining and actual writing process.
I have to publish my final draft in Word for the Papers citation manager to work. I would enjoy some alternatives here too.
Hi - here are a few suggestions for the different categories you mentioned…
PDF Converter, OCR: PDF Pen Pro 2, PDF Studio 2019 (does not look great but works really well)
Citation Manager / PDF storage: Bookends all the way, not sexy but amazing once you learn how to use it properly; Paperpile is great when using Google Docs (+ new iOS app)
Annotation: Highlights + PDF Expert + exploring Liquid Text (now on MacOS)
Knowledge and Note management: Obsidian, Roam Research, DEVONthink Pro, DevonAgent Pro
Final reviews and publication: Mellel works great, now excellent compatibility with Word; amazing if you are using Bookends; have a look at Grammarly and ProWritingAid for final polishing
Adobe suite replacement: The 3 Affinity softwares are top-notch (Design, Photo and Publisher), and no subscription.
Hi, I really enjoyed your post. I am actually creating a literature note on it. So I found it very interesting. There is something, though, I didn’t understand. When you say:
what do you mean? What do you do above the highlight notes?
Thanks for your feedback; I’m glad my post was helpful. Some of the early approaches to a digital zettelkasten had highlights in one note, then you’d create a separate literature note for each big idea. I thought it would be easier to type in the literature notes directly above the highlights on the same page, then add a back-link heading (a one-sentence summary of the note) that you could organise in your index. So, let’s say you read an article and highlighted some good points. You would create a note for this in Obsidian called something like [[H - How to take smart notes (or whatever the text is titled)]]. You would have your metadata at the top (author, date, link, full publication in APA), followed by all the highlights. The highlights would follow this structure:
[[E - One-sentence summary of idea)
L: The note rewritten in your own words with your own insights.
- H: The highlight from the original source
So the L note goes above the H note but below the E title. Then, you would slot the E titles into your index as contextually relevant, and then click it to create it as a standalone evergreen note.
In the evergreen note, I’d include the author’s surname and the date in brackets as a top level heading (e.g., # Ahrens, 2018) followed by the L note I wrote in my own words in the original highlight note. Following that I’d have a # Connections header, under which I could link to other related evergreen notes and tabs. Having them as headings meant I could link to the L note only elsewhere in the vault.
Using transclusion and the linking to headers feature, you can modify the E titles in the index to be something like: [[E - one-sentence summary of idea#Ahrens, 2018]], then looking in the preview mode you can see the L note transcluded in the index and the name of the author.
Over time, I found that transforming the notes into my own words was taking a long time and creating a backlog of notes to process, so I’m now not transforming notes until later on. When I went to use the transformed notes in my own publications, I invariably reworked them anyway, so now I’m following the same process as above but not bothering with L notes. Also, following some reflection on Nick Milo’s blog post on progressive summarisation, I now think of connections earlier in the process. The current approach is:
- Read and highlight sources (in Zotero) then create Highlight notes in Obsidian
- Add one-sentence summaries above each idea (keeping the ideas in the original author’s words)
- Think of connections for each idea and add tabs and short notes instead of the old L notes
- Organise E summaries into the index
- Create evergreen notes by clicking the E summary in your index, adding in the surname and date of the author, the note from in the author’s words, connections and reference. (I actually call these Atomic notes now instead of evergreen notes because evergreen notes are defined differently to what I have, which is still the author’s original work but split up into atomic ideas)
- Create paper outlines drawing on the E titles and transcluded atomic notes in the index and transform into your own words from there.
Not sure if this is interesting or useful so I’ll leave it there but happy to have further discussion on any aspect of my process. Seems to be working in that there’s no backlog of notes to be processed and I can spend more time reading and publishing.
These dumb threads cost money.
I’m finishing up my masters now, looking into maybe going for PhD. Recently started using Readcube Papers but now I’m wondering about the Highlights app. Papers is clunky-looking but includes a library and research tools and seems to have a solid group of developers and user base, so I’m still leaning toward that option. Highlights has a beautiful interface but I’m not sure if it works well for BOTH annotation AND finding additional articles/sources (correct me if I’m wrong, please!). Looks like maybe Highlights stores its information as metadata? Still playing with it.
I used to use Zotero all the time but Papers imported all my Zotero library references, so I’m thinking that might do the job just as well. Has anyone moved away from Papers because of some compelling reasons I might have missed? Really trying to narrow down my toolbox, but threads like this make me wonder about the best option before getting too invested.
Assuming you mean the PDF reading app for macOS and iOS, it is not designed to support literature searching/discovery. It is an excellent highlighting app, though, and plays well with note-taking as you can export annotations in Markdown format.
Thank you so much for the elaborated explanation. Instead of Zotero, I’m using Paperpile, which have built-in annotation, and highlight. Once I’ve finished reading and annotating the article, I can export them in Markdown format and paste it into the notes
For those using Zotero (or those not on Mac), I added a few tips over here on how to use your standard PDF reader similar to the Highlights app: Zotero best practices
Amazing tips here. Thanks arg
Thanks for your thorough explanation! I have been following a similar approach, based on your experiences for three days now (I know, not that long).
But processing the ~50 notes I took from a single article is taking me over 2 days now. Whereas i used to shortly summarise a finding from a paper in one sentence, which took 15 mins for a paper tops.
Do you think it will be worth the effort in the long run or am I taking the wrong approach here? I am intrigued to see whether I will start seeing a bigger picture more easily with this method but if I can read a fifth of the amount I otherwise would, I’m not sure if it will be worth it. Though perhaps these are just startup costs and it will get faster?
I am interested to hear your guys’ opinions on the matter!
Hi mate. I’m totally with you on this. In academia it’s critical to have a system that allows us to read and mine important ideas from papers into your vault as efficiently as possible. My method has continued to evolve and I’m finding it more efficient now.
In a nutshell, I’m now adding the one-sentence summaries to highlights as I’m reading (and the tags where possible). This means I don’t need to read the source more than once; instead I’m processing them as I’m reading because that’s when I discover them as important points in the first place. I then bring them into Obsidian in a single note per paper/source. I title each note Surname, date (e.g., Smith, 2018). It’ll make sense why in a moment. Each idea within the note is structured like this:
One-sentence summary of idea
Original idea in the author’s words (Reference, date, page number).
T: #tags #go #here
C: Any connections to other notes or ideas - not necessary to include for every idea but it’s useful to think of connections where possible
If you structure all the notes this way, it means you can then add the ideas straight into your index with transclusion without needing to create any additional notes (in the past I created a new evergreen note for each idea).
An example of a transcluded idea to pop into your index would be like this:
![[Smith, 2018#One-sentence summary of idea]]
This allows you to see the source and the summary of the note in edit mode and just that idea transcluded from your note page in the preview mode.
I have another approach for actually turning those ideas into publications, but this is the main approach for processing notes into my index. There may be even more efficient ways to do this. The key I think is being able to process ideas into your vault as quickly as possible while still tagging and making connections to help with later retrieval of ideas.
Since changing to this approach I’ve written a couple of book chapters with very little cognitive strain and I’m reading more than in the past (it’s addictive because every paper has the potential to be used to level up your knowledge base).
Hope this is somewhat helpful to others. The evolution will undoubtedly continue. I know there are awesome examples of how to do all kinds of things in Obsidian but all I’m really aiming for is being more productive in my academic role. The rest is all interesting but additional to my main purpose for this wonderful app.
This really seems to be a step in the direction that would work for me as well, I must say!
I did not really index (every) note yet, which may be a good idea to start and may well speed up the process.
However, upon trying your latest approach I found the resulting index to be somewhat cluttered. The main ideas are only the small title/summary, whereas the content, tags, and links below it takes up the most space.
I tried using the raw internal link (not transcluding) but here the “Smith, 2018#” was distracting me a bit. Do you not find this bothersome or have you also found a good solution for this?
What also may make the workflow less compatible is that my research is in chemistry, where I have very dense papers (virtually every sentence in a good paper expresses an important idea). As such a lot of paragraphs have to be highlighted and some paragraphs contain 3 or 4 important ideas.
Of course, I do not know in which field you are active, but I imagine fields with longer and less dense books/articles that centre more around developing arguments may work better with this system.
Perhaps I also need to loosen the atomicity principle a little, as I don’t yet see how Obsidian could work currently, as I am now expecting to require 4 days to process the paper I read last. (Although it was a great paper with a lot of useful info, making it more time-consuming).
This reference system is brilliant. Stealing the format if you don’t mind, but using it in the context of my own system.
Moved away from Papers before they were bought by Readcube. Currently using Mendeley but seriously considering the move to Zotero. Papers gets bogged up pretty fast and the library then is very resource hungry. Mendeley also provides a sturdy iPad app for reading and when I need to extract highlights I mark it up in PDFExpert. Similar to you I am interested in simplifying and consolidating my workflows. Zoteros would enable me the automatic import of highlights into Obsidian (as far as I understand until now), so that would help. What was you reason for using Papers?
Mendeley is a low-quality, reluctant to improve, adaptation of Zotero (based on some/many of its libraries afaik) and has sold its soul to the Elsevier capitalist devil—and is just a terrible app. I would really encourage you to make the transition sooner.
You can organize your PDF library using Zotero plugin ZotFile and put it on the cloud. Follow this excellent tutorial. Then, use Xodo PDF reader to read your files on your iPad right off your cloud drive. Per my experience on my Android phone, it does the whole syncing stuff for you very smoothly with OneDrive.
Xodo does not have a Mac desktop app. But in case you are on Windows, it has an awesome desktop app for you. I have had a great experience using it in tandem with OneDrive. With some black magic, whatever annotations you make on the files in Xodo are saved right away and you do not get OneDrive errors like “could not sync that file as it’s open somewhere.”
Hi @psyguy, thanks for the encouragement to move on, I especially enjoyed the religious attitude in your passionate dismissal of Mendeley. As it mostly is with regards to the devil I have some things that currently make it necessary to continue to use the app (sharing groups with my PhD and MD students). Meanwhile I will organize my escape to Zotero and check out the links an apps you mentioned. Thanks! I’ll report back.