Another idea with link types is to additionally have link weights, where stronger links (e.g. 1.0) get a multiplier to their link force on the graph view to make them appear closer together, while weaker links (e.g. 0.2) might have a negative multiplier.
The idea here is you could differentiate something that is extremely closely related vs something more tangential by just looking at the graph. This way, you could make “clusters” of closely-related information and have external links to other clusters. I would be really excited to see something like this.