How to Learn with ChatGPT and a Knowledge Graph | InfraNodus Tutorial

In this video, I will demonstrate how you can use infranodus.com to enhance your ChatGPT conversations with the help of knowledge graphs. You will be able to see the main ideas and connections between them and, most importantly, the less represented topics that you can develop further as well as the blind spots in your discourse, which can be very useful for developing new ideas.
Try it on infranodus.com
Timecodes:
0:00 How can knowledge graphs be useful for a conversation?
0:54 Starting a ChatGPT conversation in InfraNodus
2:07 Selecting the most relevant topics on the graph
5:01 Exploring a certain topic
7:32 Knowledge graph helps to steer the conversation toward your interests
7:45 Finding blind spots in the graph for prompt generation
10:48 Prompting yourself to think using AI
11:34 Ecological thinking with networks
13:20 Zooming out to see the main ideas
14:24 Revealing underlying ideas by hiding the most prominent ones

[@dancetechtv]

amazing

[@keithdow8327]

Thanks!

[@noduslabs]

Thanks a lot!

[@shawnfromportland]

Idea: an interactive infranodus tutorial covering absolutely all its features... IN infranodus itself. Load up all possible infranodus articles and documentation in infranodus itself to browse the resulting knowledge network to uncover features and workflows that weren't immediately obvious that you could do in the software, especially for new users. Similarly, once you've loaded in a knowledge body of your own topic, is it currently possible for infranodus to guide you on which suggested infranodus features to use on your data set? Can it come out and say "you've got a potential blind spot in your body of work HERE, i suggest you try A B and C to explore and improve that" Rather than the human needing to notice the blindspot first?

[@LuisFernandoValenzuela]

I guess you could have rules to emulate that behavior, but I think the current approach is to exploit the human's visual cortex + human context capabilities, since topic models are statistical aggregates of what can be gathered from the text corpus itself. Right now it is suggestive enough for a human to make sense of each potential "gap" (see "structural holes"). To me it, to automate it to the degree you want it is desirable, but seems to me you'd have to 1. trust an AI criteria on wether there is a "useful gap" or not, then 2. use it in all of the detected gaps, and then 3. it should decide which one has the most merit / potential for you (which would require more context from you).

[@noduslabs]

I like this idea of recording the user's activity and then recommending some gaps in how they could explore the tool further. Seems like a great product idea actually :)

[@railzip]

This is really good

[@lawrencium_Lr103]

Love this - can't wait for the VR version.

[@noduslabs]

Almost complete! I dreamt about it the other day. Roaming around an AR version, projecting ideas right into space in front of me. Was pretty cool :)

[@lawrencium_Lr103]

@@noduslabs I've envisioned a version that moves in time as a planner. Future events are further off in the distance and draw nearer as time passes. It's possible to move forward and backward, but it'll always return to present.

[@shawnfromportland]

🔥🔥

[@patrickbaumann8670]

Thank you! I’m a 25 year old BO from Austria, my question is - is there a possibility to use my local LLM API for Infranodus? Idealy also running locally(but that would be enterprise and I’m a new business not even having business cards yet, so enterprise is a lil too expensive for me now) If not, will there a way to implement other LLM Models through regular REST/SOAP/OPEN APIs? All the best, Infranodus looks mindblowing 🎉

[@noduslabs]

Thanks a lot for your feedback! We're definitely working on making it possible for you to implement local models.