Knowledge Graph / Graph View / 3D visual report

Ideas & Features
1

Just to collect all ideas and potential work in progress here around a visual Knowledge Graph, similar to what is becoming very popular as used by Obsidian and Logseq.

3 Likes
2

The idea is to show entity level relations in at least a two dimensional mindmap kind of way. What does currently come close to this in Fibery?

1 Like
3

Just wondering if this has been discussed before and if its in the mind of the Fibery architects…since this is like the hot feature that all personal and collective knowledge systems are implementing. It is especially important for systems like Fibery that can overwhelm first (and regular) with complexity.

It will also give the necessary edge to Fibery to improve its UX in easy navigation, which I find is actually a challenge still for daily use. In another post I mentioned that Fibery tries to combine relationship editing and relationship navigation, but a knowledge graph way of navigation perfectly fits the character and purpose of Fibery.

2 Likes
4

True. For the Second Brain we’ve build, this would be ideal. Currently you just have a bunch of data, but you can’t connect the dots. I also mentioned this here.

:arrow_down:

2 Likes
5

Possible using API’s:

  1. InfraNodus: Designed for text network analysis and visualization, it can be used to create a Knowledge Graph of Fibery entities.
  2. Graphy app: Provides interactive data visualizations, suitable for graphically representing relationships between Fibery entities.
  3. DiagramGPT: Generates diagrams from natural language, useful for visualizing relationships between Fibery entities in a unique way.
  4. Dreamspace.art: Allows for the creation of diagrams through node and edge manipulation, offering another method to create a visual Knowledge Graph.
  5. Vizionova: An idea visualization and collaboration tool that can be adapted to visualize relationships between Fibery entities.

For each tool, the integration process involves using its API to fetch graph data from Fibery, visualizing it, and embedding the resulting graph into Fibery as an iframe or a custom component.

Alternatively, libraries:
Open-source options that could be used to create a Knowledge Graph similar to those used in Obsidian and Logseq:

  • Cytoscape.js: This is a powerful open-source graph theory library that allows you to visualize complex networks. It’s highly customizable and supports user interactions.
  • Sigma.js: This is a lightweight but powerful library dedicated to graph drawing, which makes it easy to publish networks on web pages.
  • Vis.js: This is a dynamic, browser-based visualization library that provides an easy way of creating network diagrams and other types of visualizations.
  • D3.js: While not specifically a graph library, D3.js is a powerful visualization tool that can be used to create custom graph visualizations."

API’s Pros:

  • Ready-made solutions, saving development time.
  • Regularly updated and maintained by the provider.
  • Access to advanced features without needing in-depth expertise.

API’s Cons:

  • Dependence on external services, which may affect stability and performance.
  • Potential costs associated with usage.
  • Limited customization options compared to in-house development.

Libraries Pros:

  • High level of customization to fit exact needs.
  • Full control over the code, updates, and improvements.
  • No reliance on third-party services, ensuring stability.

Libraries Cons:

  • Requires more development resources and expertise.
  • Responsibility for maintenance and updates.
  • Potential learning curve if the team is not familiar with the library.
5 Likes
6

Has anyone gotten something working? I too would like an Obsidian-like ability to browse and navigate.

7

Also interested in this. Another method that I have as a rough concept is using some automation service like n8n or make.com to fetch data, put them in (for example) google sheets and send them to an AI embedding model.

Two possible ways to do that:

Run locally:

  • Embedding models are lightweight and can be run locally using Ollama or another app, on most machines.
  • No / Low costs. Just your power.
  • Needs to be accessed via selfmade API.
  • Does not run in realtime, unless you let your local system run constantly.

Run via service:

  • Constant uptime; realtime capability.
  • Quicker and more efficient as you can use state of the art models.
  • Does not cost too much.
  • Additional hosting costs.
  • Additional API costs.
  • Data is online.

As I said, this is a rough idea on how to engage this, but it is also something that will come up for me in the near future.

2 Likes