Try it on infranodus.com 0:00 What you will learn 1:24 Generating a knowledge graph from Obsidian's text file 2:16 Force atlas layout - revealing context and clusters 4:20 Modularity metrics - community structure 5:40 💡Finding gaps and connecting them to improve graph connectivity 7:20 💡 Cognitive variability method - see more on 8:11 Using AI to improve graph's connectivity 9:16 Concept (graph node) analysis 10:00 💡 Betweenness centrality - the measure of influence in a network- and the social network analogy 12:48 Connections between pages in Obsidian's vault as an example 13:44 Another demo of betweenness centrality 14:02 💡 Top relations (bigrams) - using them to discover hidden content 15:20 Using Dot Graph text representation of a graph to build a ChatGPT diagram 16:14 💡 Remove underlying ideas - cutting off the top layer of ideas to reveal what's hiding underneath 17:08 Analyzing the topical trends - timeline analysis of a graph 18:50 Where to get advanced graph metrics (main InfraNodus version)
@@noduslabs If I had to answer for him. I think he means on availability not behind a paywall. A slightly generous free tier would be great. Also would look into what Amplenote does by offering the paid version to people that actively promote their product.
Hey Dmitry, this looks great! I use Obsidian a lot, so I guess this is why RUclips recommended me this video. Do you have plans for something less automated? My use case is very different, but I keep hoping to see someone implement a professional graph analysis solution for it (right now I have a mish mash of my own tools, formats and pipelines). With your examples, it's all very dynamic, I assume you're targeting initial analysis, or insight extraction, or ideation, or other statistical processing, something like that (I'm not sure, so please forgive me if I'm wrong). In my workflows this is a very rare situation, for a very brief time interval, just orienting myself around some body of textual data (maybe a set of papers or smth). Most of the time, I need to have stable, editable, versioned views. I don't want nodes to jump around, I want to remember where I left them the last time. I don't want anything extracted randomly, I want to freeze some part and augment with some new subgraph. And so on. I want to be able to manually add/remove/edit subgraphs. Are you going to have functionality like this?
Thank you for this feedback! Well, actually the nodes' positions look exactly the same every time you open your Obsidian files. So you can always orient yourself around your content. Same goes for the folder views. You can just set up to look at the [[wiki links]] only, so the view will be quite stable. When you're talking about "freezing the graph" and augmenting it with another one, could you please elaborate what you mean, so I can understand it better? Thank you!
@@noduslabs thank you! I need to take a close look then. When I talk labout freezing/augmenting the graph, I mean operating on it semantically, like you'd operate on a diagram, by directly injecting/altering/moving the spatial representation. Say, I have crawled a folder of latest papers on some topic, now I want to merge concepts based on rules, I want to version graph views, not edit in place, I want to have filtered representations and switch between them, e.g. layer the graph, not see it as a cloud. Then I want to come back to this graph later, when I read a new paper on the topic, and integrata that paper into the existing graph. Basically, my problem that I'm trying to solve (for years, lol) is semantic operation on graphs as first class sitizens, not graphs as a derivative from textual or statistical substrate. I hope it makes sense? But then again, I need to check what you have right now, as YT just recommended me this video and that's all I have seen. Thanks for your work!
@@matveyshishov6339 Hey, thanks for the explanation! Yes, in fact, we will introduce these features into the next versions of the plugin based on the users' feedback. You can already do that in the main version of InfraNodus on infranodus.com - to create the nodes you just use the text panel on the bottom left (you have to type the name of the node anyway) using the [[syntax]]. If you want to make a connection, you use [[object_1]] to [[object_2]] and describe the connection. It's much faster than any visual interface I found. What you need, I think, is that when you do that on the Obsidian plugin itself, it creates the according pages inside your Obsidian vault, so you can edit the structures using the graph and then add textual content when you're done building the structure, right?
The tool is called InfraNodus. The knowledge graph is created on the fly by InfraNodus. The plugin is called InfraNodus. You can find it on infranodus.com and the plugin is available on infranodus.com/obsidian-plugin
I am thinking about how to visualise the literature review matrix using this tool to categorise the themes and identify the gaps in the literature. Can you make a video on it.
![How to chat with your PDFs using local Large Language Models [Ollama RAG]](http://i.ytimg.com/vi/ztBJqzBU5kc/mqdefault.jpg)
Try it on infranodus.com
0:00 What you will learn
1:24 Generating a knowledge graph from Obsidian's text file
2:16 Force atlas layout - revealing context and clusters
4:20 Modularity metrics - community structure
5:40 💡Finding gaps and connecting them to improve graph connectivity
7:20 💡 Cognitive variability method - see more on
8:11 Using AI to improve graph's connectivity
9:16 Concept (graph node) analysis
10:00 💡 Betweenness centrality - the measure of influence in a network- and the social network analogy
12:48 Connections between pages in Obsidian's vault as an example
13:44 Another demo of betweenness centrality
14:02 💡 Top relations (bigrams) - using them to discover hidden content
15:20 Using Dot Graph text representation of a graph to build a ChatGPT diagram
16:14 💡 Remove underlying ideas - cutting off the top layer of ideas to reveal what's hiding underneath
17:08 Analyzing the topical trends - timeline analysis of a graph
18:50 Where to get advanced graph metrics (main InfraNodus version)
Amazing cutting edge application! I look forward to a time when it is more widely available to a larger community.
Thanks! What do you mean by more widely available by the way?
@@noduslabs - More accessible to more users.
@@noduslabs If I had to answer for him. I think he means on availability not behind a paywall. A slightly generous free tier would be great. Also would look into what Amplenote does by offering the paid version to people that actively promote their product.
@noduslabs of course nothing beats open source as well
@@noduslabsHe means free basic version
This is a wonderful tool for research and education! Bravo! 😎🤖
Thank you! Glad you like it!
Your best video, yet!!!
This absolutely incredible. Can’t wait to try and spin this up.
@@kevon217 great! Here’s the plugin: infranodus.com/obsidian-plugin
Please let me know what you think!
Hey Dmitry, this looks great! I use Obsidian a lot, so I guess this is why RUclips recommended me this video.
Do you have plans for something less automated?
My use case is very different, but I keep hoping to see someone implement a professional graph analysis solution for it (right now I have a mish mash of my own tools, formats and pipelines).
With your examples, it's all very dynamic, I assume you're targeting initial analysis, or insight extraction, or ideation, or other statistical processing, something like that (I'm not sure, so please forgive me if I'm wrong).
In my workflows this is a very rare situation, for a very brief time interval, just orienting myself around some body of textual data (maybe a set of papers or smth).
Most of the time, I need to have stable, editable, versioned views. I don't want nodes to jump around, I want to remember where I left them the last time. I don't want anything extracted randomly, I want to freeze some part and augment with some new subgraph. And so on. I want to be able to manually add/remove/edit subgraphs.
Are you going to have functionality like this?
Thank you for this feedback! Well, actually the nodes' positions look exactly the same every time you open your Obsidian files. So you can always orient yourself around your content. Same goes for the folder views. You can just set up to look at the [[wiki links]] only, so the view will be quite stable.
When you're talking about "freezing the graph" and augmenting it with another one, could you please elaborate what you mean, so I can understand it better? Thank you!
@@noduslabs thank you! I need to take a close look then.
When I talk labout freezing/augmenting the graph, I mean operating on it semantically, like you'd operate on a diagram, by directly injecting/altering/moving the spatial representation. Say, I have crawled a folder of latest papers on some topic, now I want to merge concepts based on rules, I want to version graph views, not edit in place, I want to have filtered representations and switch between them, e.g. layer the graph, not see it as a cloud. Then I want to come back to this graph later, when I read a new paper on the topic, and integrata that paper into the existing graph.
Basically, my problem that I'm trying to solve (for years, lol) is semantic operation on graphs as first class sitizens, not graphs as a derivative from textual or statistical substrate.
I hope it makes sense?
But then again, I need to check what you have right now, as YT just recommended me this video and that's all I have seen.
Thanks for your work!
@@matveyshishov6339 Hey, thanks for the explanation! Yes, in fact, we will introduce these features into the next versions of the plugin based on the users' feedback. You can already do that in the main version of InfraNodus on infranodus.com - to create the nodes you just use the text panel on the bottom left (you have to type the name of the node anyway) using the [[syntax]]. If you want to make a connection, you use [[object_1]] to [[object_2]] and describe the connection. It's much faster than any visual interface I found.
What you need, I think, is that when you do that on the Obsidian plugin itself, it creates the according pages inside your Obsidian vault, so you can edit the structures using the graph and then add textual content when you're done building the structure, right?
@@noduslabs Awesome, thank you very much! I'll play with it this week.
very nice
Wich video is this with the introduction in InfraNodus, so that I can understand how I need to work with it and to know what capabilities it has.
This one is a pretty good intro in a similar direction: ruclips.net/video/XkVtGS-v_7k/видео.htmlsi=_9EPehVSZwssrd_z
Also this is an intro to InfraNodus interface: ruclips.net/video/vOG9XBStUr8/видео.htmlsi=uWjBCk-BLo0v8NVe
@@noduslabs thank you
@@noduslabs thank you
Which visualization tool is used and which knowledge graph database is used internally?
The tool is called InfraNodus. The knowledge graph is created on the fly by InfraNodus. The plugin is called InfraNodus. You can find it on infranodus.com and the plugin is available on infranodus.com/obsidian-plugin
I am thinking about how to visualise the literature review matrix using this tool to categorise the themes and identify the gaps in the literature. Can you make a video on it.
Sure can you please give an example of how the data looks like? Maybe some sample files?
I wonder How to get an jnfranodus graph from Evernote notes or tags?
We have an integrated Evernote import in infranodus.com - you just connect your Evernote account, choose the notebook, and analyze it.
Obsidian
What is?