This repo contains NLP analytics on some of my Obsidian.md vaults; I am experimenting with how algorithms can improve knowledge management workflows, via the functionality of Obsidiantools and NLP libraries.
The main focus of the NLP analytics is topic modelling.
These are examples of functionality for which I am exploring proofs of concept:
- Auto-generated Maps of Content (MOCs)
- 📓 Notebook: proof of concept for auto-generation of MOCs for my MPhil vault.
- Auto-suggested wikilinks
- 📓 Notebook: proof of concept for auto-suggested wikilinks for my film noir vault.
My vault content is private. However the information in the notebooks and scripts that I have written could be re-applied to your own vaults with a little reworking.
These are vaults that I am using for prototyping:
- 🎓 My MPhil degree notes
- For my master's degree I wrote all my notes in an Obsidian.md vault.
- The vault has over 100k words of content spread across 500 MD files. I didn't strictly follow the use of atomic notes. Each note had a concept or keyword, but the notes varied in length and were optimised for ease of revision. The typical note could fit on an A4 page (a few hundred words of content). I essentially used the vault as a wiki.
- I studied 11 modules, so I assigned notes to folders for their respective modules. I averaged about 50 notes per module.
- The rigid structure of folders was convenient for revision. I didn't develop any MOCs as the content was manageable for revision on a module-by-module basis. However, one MOC per module would be great for future reference. I am interested in auto-generating MOCs from the vault content and exploring which notes are the most important for each module.
- 🎥 Film noir vault
- I created a vault on 100+ film noir movies from IMDB data.
- There are a few common themes in film noir so it would be interesting to see how films could be connected by their themes. Examples of themes include murder, love, fear, corruption, etc.
- I couldn't possibly develop a vault with notes on 1000s of films manually, so solutions that can create MOCs and/or wikilinks automatically would be great. This could help me to discover new films to watch.