Can I just say that this is a space I’ve been deeply interested in for a long long time (I have a terrible memory, and latched on to tech early), and it blows my mind that very few (if any) of these solutions use a graph database as the back-end.
I understand the argument that it’s likely overkill for <10000 entries, but for me it’s been wildly freeing. Especially in cases where the abstraction layer gets out of the way and allows for free-form connections between items. No more rigid db schemas, primary keys, or trying to cram “one or many” connections in to a column.
I think I come at this problem the opposite way to everyone that actually builds these products/tools. I want to mindmap and brainstorm all of the topics, and then take notes on them - preferably with a time component to that. Even more than that, I want to be able to take notes on the relationships between nodes, since that's where the real gold tends to be.
Mindmap tools generally don't work for this because they want a single topic for each map, and all the words on the one page. KM/KB tools don't work because I can't just draw the relationships between the points, I have to go in and actually create notes that I then get to visualise.
I really wanted to like Obsidian, but the way it treats folders puts me right off. I tried Neo4j Bloom, but there are UI difficulties and it's not really great for notes. I'd build it myself but I have no idea where I'd even start on a UI.
I'm the same way. Mindmaps are by far the best tooling for how I think. They're also just not portable. I keep trying. It would be awesome if mind maps could be united with a console somehow, so that I could inject things from "wherever".
I've spent a lot of time caring about KM/KB and can't settle on anything that survives more than 2 years. What do you do today?
I once built a mind map with a “MUSH” interface (text-based, where you traverse the map “on ground level” as a character).
You avoid information overload by showing the synopsis as you “enter” a node, then you could say `e[xamine] here` to view the whole text.
Traversal is quicker than you’d think, and it would likely work well on mobile devices.
It was years ago, but I still think about making that in to something “real”.
I keep everything in a combination of OneNote and Google Sheets, which I link back to Salesforce with the Salesforce connector for Google Sheets. It is... not perfect, but spreadsheets were the only way I could find to quickly get the information down that I wanted to.
This is the combination of things that originally led to my loving TagSpaces. I wanted local files that I could edit, have folders of varying file types, and include documents in the folders, opening as needed. Part of the issue was that I then wanted to link to, and from varying files... basically a knowledge base that was independent on data types.
Turns out that image capture, and document sync was the least of my problems. The mind maps, image data not being OCRed, and other things became a pain.
I actually considered using a graph database for our highly connected notes app[1], but in the end went with PostgreSQL because I was following the advice of "go with the technology you know".
It would definitely be interesting to go back and try something like Neo4j, but we've actually managed to wrangle postgres for our graph while still maintaining pretty good performance.
I understand the argument that it’s likely overkill for <10000 entries, but for me it’s been wildly freeing. Especially in cases where the abstraction layer gets out of the way and allows for free-form connections between items. No more rigid db schemas, primary keys, or trying to cram “one or many” connections in to a column.
Shoutout to Neo4j and APOC, as well as to one of my inspirations: https://neo4j.com/blog/building-graph-history-codex/