a list for 'against comparison'
a list for 'against comparison'


Generative Insight with AI and Network Thinking.

What is Infranodus

claim there software gives you an overview into any discourse, revealing the blind spots and enhancing your perspective.
It can vectorises small data sets (CSV, PDF’s, Docs etc) - and is best suited to tablictour data
There use-cases include scientific & non-profit research.
Screenshot 2024-02-13 at 08.30.30.png
In this research, I am exploring the feasibility of using Infranodus for Production Library Publishers:
Indentify gaps in production library catalogues
can it compare and identify blind spots between different publishing catalogues
some user advice
legals and data

Find Blind Spots

Screenshot 2024-02-11 at 14.00.48.png

Preparing Data

Infradanous can be a transient and repetitive process - disposable, like a ChatGPT thread.
It excels at ‘anecdotal’ discovery and you’ll likely starting a fresh, over and over again with the same data-set.
Screenshot 2024-02-12 at 16.10.15.png
End-users with research and journaling abilities will benefit the most.
It can be hard to quantify its output into a tangible actionable data points without foresight into what you want to measure as preparing the CSV in acordingly
There are essential two ‘flavours’ of filters - for ‘graph connectivity’, how the nodes are connected to each other, selected nodes can either be highlighted or hidden - I’m a big fan of hiding nodes, as it helps with performance.
Screenshot 2024-02-13 at 18.36.29.png
Screenshot 2024-02-13 at 18.36.58.png
The actual ‘Filter’ is essential your ‘search result’ a list for against comparison. for what you want compare the nodes against - you don’t want too many of them
Screenshot 2024-02-13 at 19.28.10.png

Screenshot 2024-02-13 at 18.49.46.png

Graph Connectivity vs. Community Structure

Graph Connectivity refers to how nodes (entities in the music metadata graph) are connected to one another. A graph is fully connected if there is a path from any node to every other node in the graph. High connectivity in a music metadata graph might indicate a genre where artists frequently collaborate or where there are strong stylistic overlaps between songs and artists.
High Connectivity: May suggest a well-established genre with many collaborations.
Low Connectivity: Indicates isolated nodes or clusters, suggesting niche genres or less common collaborations.
Community Structure in a graph refers to the formation of clusters or groups of nodes that are more densely connected to each other than to nodes outside the group. In music metadata, these communities could represent genres, subgenres, or even geographical scenes. Identifying community structure is crucial for understanding the landscape of music genres and the relationships between artists within and across these genres.
Distinct Communities: Reflects clear genre or scene distinctions with strong internal similarities.
Overlap between Communities: May suggest emerging genres, crossover artists, or the fusion of musical styles.

Structural Gaps in Data :

Here is some research into using Infranodus to find structual gaps in production music library data.

Screenshot 2024-02-11 at 08.06.10.png

Okay. This looks amazing. I’m going heads in first to generate some data to show the boss. What do I need to know?
One innovative idea that could link these concepts is a reality television show called "Drive to the Stars." The premise of the show is to find the next big musical sensation by putting a twist on traditional talent competitions. Instead of holding auditions in a studio, contestants must perform while driving a car.

Infranodus excels at ‘anecdotal’ discovery.
It can find gaps - but are they all worth filling?
managing the context window of this product can be challenging

anyone trying this needs to have enthusiasm with ‘large language models’.
You’ll need to learn the idosynciaes of how Infranodus manages its memory, as you’ll be resetting the nodes a lot

You need to supply it CSV sample sets of AI auto-tagged music metadata. This will need to be a uniform standard - ie AIMS, Cyanite.. - of all the data.
If you require audio playback, your sample data will have to include url-refs to an accessible CMS.
When you import CSVs, you are limited in how many columns you can filter - don’t go overboard on your first time with huge multiples.
my recommendation to start would be composer and song key, as they are both fixed with multiples, they lend themselve well to more ‘scinetific use-case nature of this.
Infradanous is a transient and repetitive process - you’ll be starting fresh over and over again - so end-users with strong research and journaling abilities will benefit the most.


All your input needs to be from a single source of contextual AI tagging ie - AIMS, or a Cynanite... Consequently customers for this product should look for a self-hosted solution to mitigate limitability about using there AI tagged outputs to train other big-data.
Catalogue Size
At the moment, there is a 3,800 record limit and comparing one vectorised data set to another (i.e a sub published catalogue) would likely require having to build some sort of caching mechanism for data - as using an LLM to translate between the two data sets would just end up incurate, synazited data.
Screenshot 2024-02-13 at 13.41.37.png

Want to print your doc?
This is not the way.
Try clicking the ⋯ next to your doc name or using a keyboard shortcut (
) instead.