A Visual Strategy to Organize Academic Reading

As academics, we don’t have a shortage of papers to read. The challenge is firstly to understand what is important and why. Secondly, to capture it in a meaningful and actionable way, not just adding it to an endless “to-read” list. Over the years, I developed a method to capture papers in a visual way: The reading map. Using AI, this process is now as easy as clicking the “Add to Zotero” button. The result is an organized network of papers with priorities, topics and relations to one another. I know exactly what is worth my time to read once I sit down. Let me walk you through this method.

Here is a bird’s-eye view of the reading map to give you an idea of what I mean. Each (unread) paper is stored as a card with title, abstract, link and a core question that it answers. The map is structured by topics, and relations between papers are represented by arrows.

For example, one paper “A” might present an important method. Another paper “B” evaluates this method and compares it to a global benchmark. Paper “C” is an application of this method. The map stores this as A→ B→ C, because the logical order of reading these papers would be from A to B to C. Similarly, if two papers contradict each other, they should be read together.

This is what the reading map allows you to do: Store relationships between papers and reasons to read them so that you do your reading in context. Additionally, it helps you not to be overwhelmed, as you are only dealing with a subset of the map, even if the map contains dozens of pages.

Here is how to create such a map in Obsidian, using the canvas feature. Canvas is a versatile feature I use for many other topics like managing projects. If you have never used Obsidian before, check out why I think Obsidian is the best note-taking app for academics.

Step 1: Finding Papers

The easiest way to find important and impactful papers is by starting from just a single seed paper. Simply googling your keywords will produce countless papers. Pick just one that seems relevant. Keep in mind that Google Scholar is biased towards older, more cited papers and is therefore a limited option for a comprehensive literature review. Instead, upload your seed paper to Litmaps or another dedicated literature discovery tool. Here is a comparison of three very popular ones:

These tools allow you to build a collection of papers by using the reference network between the papers, which leads to better, more relevant results. My literature review course explains this “detective approach” in depth. But regardless of how you search for papers, you end up on a website that shows your paper, its abstract and authors. This is what we are going to need for the next steps.

Step 2: Creating the Canvas

Canvas is a core feature of Obsidian, the note-taking app for academics. It is fairly simple to use. Create a canvas by clicking the canvas button in the left side bar. The canvas has three elements:

1. Cards – Just some text that can be formatted like a note but lives inside the canvas
2. Note – A note stored separately but displayed in the canvas
3. Image – Just an image

In addition, you can connect the cards notes and images and add some basic styling:

That’s it. It usually only takes a few minutes to get used to using the canvas, but the applications for researchers are countless. So let’s get started on creating our reading map.

Step 2: The Paper Card

To organize our reading map, we are going to use cards to store the title, abstract, link and key question for every paper we intend to read on a canvas card. It looks like this:

Cards can use the markdown format to add links, headers, bold or italic formatting. Depending on your Obsidian theme, this will look different for you. I prefer using the “Border” theme because it assigns different colors to bold and italic formatting as well as the headers.

With this structure in mind, you can start creating cards for the papers you found. The key here is to organize them in a contextual layout that makes it easy to recognize groups of papers or logical sequences.

Step 4: Connecting Cards Logically

The most important part of your reading map lies in its connections. Each paper logically fits into your learning goal. First, you learn the basics of a topic, then you go to specific subtopics or applications and so on. You can also have multiple topics and sub-topic in your reading map. This creates a network of dependencies that are stored as connections between the cards.

In the diagram above you can see that the central topic is the “Trait – Environment – Space” relationship (an important concept of my research in ecology). There are sub-areas, for example, the relation of traits to ecosystem functioning which is described in the paper to the left. That paper is more generalist and another paper (bottom left) is a specific, narrower case-study of this. By connecting Topic → Paper → Specific Paper… I built a narrative that can help me decide which papers to read in which order.

The questions I stored on each paper can further help me decide which paper is the one I will be reading next. It tells me how to learn my domain step-by-step without feeling overwhelmed by the 100s of papers that might be relevant. Compare this structure to a list with 100s of papers that gives you no clue where to begin or end.

Step 5: Saving Research Papers using AI

When it comes to building to-do lists and reading maps the key is to make the process frictionless and easy. If you it takes you minutes to add a single paper, you will likely not do it and instead click the “Add to Zotero” button in your browser and add it to a list. Luckily, AI can help you create the paper cards from a screenshot of a paper. This is how a paper looks if you find it on a website or on Litmaps.

Notice that the screenshots contain all the metadata you will need. To extract this automatically into a canvas card, we will build a Custom GPT assistant. ChatGPT assistants are bots that have a set of instructions and can use ChatGPT to, for example, extract information from a screenshot. Here is a tutorial on building Custom GPTs. We are going to give our custom GPT these instructions:

I will prvide you with a screenshot. 
Your task: 
0. Check if I added a question with the query. If I didn't add a question, formulate a single short question that is being answered by this paper and use this
1. Extract the first author + year
2. Extract title 
3. Extract DOI
4. Extract Abstract
5. Combine them into this template
<template>
```markdown**{Question}**
### [{First Author} {Year}](https://doi.org/{doi}) - {title}
{abstract}```
</template>
Always print out the result in a separate box with a copy button.

The format here is a little bit cryptic if you are not used to coding. In a nutshell:

• Curly brackets are used as variables: {Name of paper}
• <template> tells the AI that this is a template and has to be used without modification
• “`markdown is a keyword that enforces ChatGPT to use a markdown format box.

Let’s test our assistant and copy and paste the answer into the canvas:

The only thing you need to do is to screenshot the webpage, drag the screenshot to your GPT and copy the output to Obsidian. It takes only seconds. If you want to learn everything about building intricate Custom GPT assistants that can save you countless hours of time, check out the Effortless AI course.

Step 4: Reading Papers

The final step is to read the papers – the fun part! After reading the papers I remove the old card and replace it with a note that contains all my notes on this paper as well as links to other concepts mentioned therein. All of this is part of the Source-Topic-Argument note-taking system for organizing your information.

After reading most of your papers, your reading map turns into a comprehensive reference map. The advantage here is that to remember a paper you can now leverage your spatial memory. Since you don’t just remember the title or author, but the location on the reference map. Our brains are excellent for remembering spatial arrangements; most memory techniques leverage this fact. Now you can easily remember not only the logical relations between papers but also easily find them in your vault. Here is what a finished reading map looks like and how easy it is to access PDFs and navigate around the map:

Academic Knowledge Management

This workflow is part of what I call “Academic Knowledge Management“. This is a set of strategies to keep an overview in a massive set of notes, papers, ideas and research projects. If you have never heard of it, check out my AKM course where I go into numerous techniques like the one presented in this article:

Summary – How to organize a collection of research papers?

The reading map lays out all your papers on a 2D Canvas using Obsidian. Each paper is a card containing the abstract and title as well as your reading intention or question. Connections between cards/papers indicate logical relations between them and an order in which to read. Papers are clustered around topics to strengthen memory and always make the best decision on what to read next.

The beauty of the reading map vs a reading list approach is that you have context. A list overwhelms you with too many entries, a map tells you where to start and where to go next and why. Aside from this, it helps you develop a narrative for your discussion section as these A→ B→ C connections in the reading map can become sentences for your discussion. For instance, “The problem of C has been solved using B, which overcame a previous limitation of A, enabling …“. The reading map is therefore not only a tool for storing your papers, but also a tool for synthesizing them (effortlessly, so to speak!).