Finding Research Gaps in Academic Zettelkasten
An academic zettelkasten requires additional meta-level work beyond the knowledge accumulated through atomic note creation. A researcher should track the state of the literature review and stay on top of the knowledge material so that research gaps emerge during zettling without effort.
Right now, I create a dedicated source note including the freewriting I do to identify what will go out from the paper (I need freewriting to decide on material that will be created from the paper) and highlights.
Als,o I don't read papers end-to-end mostly because I am new to the field I am working on. I start with reviews, dissect reviews into subproblems, and use a sandwich technique. Let me give an example.
I am reading a paper about how spatial navigation may be the intersection point towards the efforts to merge neuroscience with AI. The paper talks about the neurobiology of spatial navigation, lists major models from different schools of thought, e.g., attractor models, reinforcement learning-based models, and deep learning-based models.
I started with this paper because, as a newbie researcher, I aim to design biologically inspired algorithms and models, but I am new to the field of spatial navigation.
So after reading the neurobiology part, I paused and searched for the major neurobiology work. The main problem of the field is to understand path integration (how we use internal cues to track our path without reference to any external input)
There are 3 major cell types occurring in path integration, namely place cells, grid cells and head direction cells. Besides these, there are many intermediary cells that may intermix some variables from the cells I mentioned at first.
I am treating this part as a black box because I need to divide the path integration problem into major cell agents, understand them in detail, and then examine how they are connected in circuits.
And there are models. This is the high-ROI part of my question.
Models start from a primitive understanding, and based on the requirements, you add or change stuff.
Right now, I am keeping a "Spatial Navigation Models" canvas in my Obsidian to see existing attempts, how they have evolved, and how they are connected.
So far, I mentioned my source note (for highlights, freewriting, and keeping track of what has been read) and canvases.
I envision these types of canvases as the places where reference/research gaps will emerge. My dream is to create a web of models, and through time, the models that are not already connected will be novel contributions.
What do you think? What do you do to keep track of the literature? If you are an academic, what's your workflow in finding research questions?
Selen. Psychology freak.
“You cannot buy the revolution. You cannot make the revolution. You can only be the revolution. It is in your spirit, or it is nowhere.”
― Ursula K. Le Guin
Howdy, Stranger!
Comments
You may find the following articles interesting:
https://effortlessacademic.com/a-visual-strategy-for-finding-research-gaps/
https://effortlessacademic.com/find-research-gaps-automatically-through-reference-gaps/
My problem with EffortlessAcademic is that he doesn't use the principle of atomicity as much as folks here.
I am wondering how the diagrams he created here can be embedded to the atomic notes.
In the same way, probably. When you have the atomic notes at hand, creating these diagrams is a no brainer.
However the problem is creating those atoms. I try to follow a strict ontology and wanted to ask what other people do.
Selen. Psychology freak.
“You cannot buy the revolution. You cannot make the revolution. You can only be the revolution. It is in your spirit, or it is nowhere.”
― Ursula K. Le Guin
Some of us venture beyond a purist Zettelkasten. ;-)
If you like atomicity, you might find inspiration in Joel Chan's work.
I find the article Knowledge synthesis: A conceptual model and practical guide interesting. It differentiates:
In the article Discourse Graphs for Augmented Knowledge Synthesis: What and Why (PDF) Chan introduces the term discourse graph. The basic idea isn't that novel, but I like the term. There is also a forum discussion about discourse graphs: https://forum.zettelkasten.de/discussion/2509/discourse-graph-and-zettelkasten.
You might also like Joel Chan's Working Notes written in Obsidian and published with Obsidian Publish. They distinguish various note types:
Then there are simplified methods for note-taking and critical thinking:
I mess up everything when I don't have a structure to follow because I haven't even started my masters and not really knowledgeable about the scientific method when bombarded with review papers
Here's the ontology I settled on based on my field with the help of ChatGPT. I think it will provide a rich experience for learning.
And I don't think "evidence/argument" bla bla really encapsulates Sascha's building blocks idea.
ChatGPT suggested to add orthogonal dimensions for every type of entity which will be used as tags. Query-making will be easier. Next step is to integrate Popperian methodology into querying the notes.
Ontology
Core causal / epistemic
explains(Mechanism → Phenomenon)
induces(Mechanism → Phenomenon)
supports(Evidence → Mechanism)
supports(Evidence → Phenomenon)
falsifies(Evidence → Mechanism)
Measurement layer
measures(Method → Phenomenon)
operationalizes(Method → Phenomenon)
tested_by(Mechanism → Method)
Mechanism structure
constrained_by(Mechanism → Constraint)
requires(Mechanism → Constraint)
realizes(Mechanism → Mechanism)
approximates(Mechanism → Mechanism)
Conflict layer
conflicts(Evidence ↔ Evidence)
contradicts(Mechanism ↔ Mechanism)
generates(Evidence → Tension)
Cross-level / mapping
maps_to(Phenomenon ↔ Phenomenon)
depends_on(Phenomenon → Constraint)
I also created taxonomies (thanks ChatGPT for everything!!)
Taxonomy of mechanisms for example:
I don't know if this much endeavor is even necessary tbh
But working with templates is an easy approach for learning, imo
Selen. Psychology freak.
“You cannot buy the revolution. You cannot make the revolution. You can only be the revolution. It is in your spirit, or it is nowhere.”
― Ursula K. Le Guin
Have you considered reading a textbook or taking an introductory class about the scientific method, that teaches the basics on how to read and evaluate such papers?
Good luck!
Right now I am reading "Criticism and the Growth of Knowledge" by Lakatos
Selen. Psychology freak.
“You cannot buy the revolution. You cannot make the revolution. You can only be the revolution. It is in your spirit, or it is nowhere.”
― Ursula K. Le Guin
@harr said:
I would second @harr's suggestion to look at Joel Chan's work, especially "Discourse graphs for augmented knowledge synthesis: what and why", cited above. By the way, in this paper, Chan cites a great paper, Hars (2001),1 that shows how Hars synthesized his own conceptual model (ontology) of scientific knowledge from existing models. You may find it helpful.
In my understanding, a "research gap" can be stated as a research question. This is explicit in the first image of the first article cited by @saf_dmitry above, where the example of a research gap is stated as a question. It's no surprise that Joel Chan and I and many others use question notes in our note systems. Even Hars (2001) has a problem element in his conceptual model. (Though I think one could differentiate between problems and questions, like Booth et al. do in The Craft of Research, the two are closely related, and I don't differentiate them in my note system.) You have been advised!
EDIT: Here's a relevant quotation from the conclusion of Hars (2001):
Alexander Hars (2001). "Designing scientific knowledge infrastructures: the contribution of epistemology". Information Systems Frontiers, 3(1), 63–73. ↩︎