Research and writing using knowledge graphs

gimalay

I was working on an experimental analysis of the philosophical texts using LLM's and very impressed on how well LLM's work in a combination with the knowledge graphs.

I was able to discover connections between different texts and synthesize a fascinating article on top of the source material. The article itself is deeply connected to the nodes of the knowledge graph allowing to see source material and explore more. The article is here:

https://iwe.pub/seventeen-centuries/virtue-across-centuries/

I wonder if you experimented with something like this and what was your experience.

If you want to learn more about how this graph was constructed all the sources are available here:

https://github.com/iwe-org/seventeen-centuries

harr

For readers who are interested in the topic, here's a related post:

https://forum.zettelkasten.de/discussion/3442/anyone-using-ai-agents-with-their-zettelkasten-thoughts-on-graph-access

Andy

@gimalay: Technically, this project is very impressive. In a word: cool!

But philosophically, as I have revealed about myself elsewhere, I am not a big fan of Nietzsche, so I am not a big fan of the fact that this project culminates in Nietzsche, as if he is the end of history, a century before Francis Fukuyama. It feels very nineteenth-century, as if the author is a a kind of Rip Van Winkle who falls asleep in 1890 and wakes up in 2026 and spontaneously discourses on "Virtue across seventeen centuries" with no knowledge of the twentieth century.

To me, the philosophical result feels scholastic in a pejorative sense, like a scholar studying a certain period in the history of philosophy who can't see beyond the boundaries of her scholastic specialization.

I assume all the text in the project was generated by the LLM except the source text?

gimalay

Most of the text is generated, yes. You can find the details of how it’s generated in the repository readme.

The article is written in collaboration with LLM, I found the difference in the perspective on the same concept fascinating and wanted to explore this more.

c4lvorias

I am doing something similar here with ontologies in an academic setting.

https://forum.zettelkasten.de/discussion/3451/finding-research-gaps-in-academic-zettelkasten#latest

Besides ontology I mentioned in my own post, I have also queries that I am planning to use alongside LLMs (which are also created by an LLM. I heavily use them for augmentation.)

https://share.note.sx/9kjsu7hk#jiUT8iZNsoT97N3l5jmjKZUjf9hRhfwFuaBUx9aDnzc