I've considered this for some time. Ideally, it should incorporate connections and context between those connections and at locations. The best technology for this is graph databases and, by implication, a standards based graph database. These are also known as Knowledge Graphs.

This directed me to the Semantic Web, RDF, and OWL. There are many commercial, open source, or otherwise free solutions using this technology. I'm using Fuseki. It's interesting to note that the USGS has a working group on Semantic Web and works to incorporate and expand geolocation ontologies for mapping.

Additionally, you can use your own ontologies and mix-n-match as needed. Most of these databases also provide for powerful inferencing to help in your queries. You could link map files and/or store drawable data directly in the stores. Also, the tech allows for parallel queries across multiple data stores simultaneously and can collate results.