/tech/ - Technology★

The hardest part would be building the connections because you would have to do that yourself because AI is really bad at figuring out things like that from text, and good luck doing that for images and other mediums. But as far as a software that at least simplifies this process goes, that'd be easy to implement. I just hope you have enough autism power to build a big enough database for what you need to do.

>>966026 (OP)

A monkey could develop a simple version of this.

>>966026 (OP)

>arbitrary relationships

>some measure of quality of relationship

>data linked to the relationship

Does not look easy to me.

For visualizing, have you tried mindmaps?

>>966027

I don't intend to use AI to find the links, at least not any time soon. Users would manually be putting them in as they find them. Generally this would be single user, but it would be interesting to allow groups to work on the same topics.

I've got about 20 years of programming experience so I'm not worried about implementing it, I really was hoping to find similarly motivated devs, or prior art to mimic and build off of.

>>966031

The system need not be intelligent itself. I think it's best modeled as a graphdb where the edges can also contain metadata such as type of link and corroborating data (text/image/audio etc...)

>>966035

I've seen them but never used them much. IMHO they get too complicated too quickly because all the edges are in the same "namespace". I want to turn connections on and off based on what I'm looking for. Maybe the mindmap paradigm is write, it just needs to be dynamic?

>>966036

So you want something like where the edges have some metadata attached to them that allows you to filter them? Like for example, you gave the example of asking "who is connected to XYZ org?" so you'd specify a filter on People, and all the links connected to XYZ org of the People category would show up, but hide everything else? But then, you'd want to have a filter for how deep the connections go. Do you want to find connections to the 2nd degree, 3rd degree, etc.? So if you wanted to find CEOs of companies Z who has some nebulous connection to company X through other companies/orgs/events/text/etc. Y, you'd have to filter on CEOs to the 2nd, 3rd, 4th, 5th, etc. degrees. Might also need exclusion filters...

>>966026 (OP)

The software you're looking for is Maltego

This is pretty much what search engines do right?

>>966088

Yeah Google's PageRank (at least at the beginning), did something like this at it's simplest layer, they also index a ton of text, and spend a ton on crawlers.

>>>/pol/11970419

>>/pol/11970419