Conversation

If you think of a plain old physical map, say the map of a territory, you can imagine a function that transforms it into a graph: - Every entity of interest in the map is a node. The node's location is the entity's location. - Edges can be links to the nearest nodes.
1
Edges can be assigned through other principles; they are just relations that are interesting. If you want to relate more than two things with the same relation, that makes it a hypergraph.
2
1
I'm no expert :) But: They are a generalization of graphs: graphs have nodes connected by lines, in such a way that a line goes from precisely one node to precisely another. Hypergraphs are more like a set of subsets: they are good for modeling tertiary or higher relations.
1
Applications of hypergraphs are actually relatively rare, I think? At least I don't remember covering them in algorithms and data structures classes, whereas simple graphs are all over the place (they are very useful).
1
These are the weak recent references I have about them in my cache :) (If you are famous and I @ you: please excuse the intrusion, I just try to include platform references when referring to well known entities, and that's @-ing people on Twitter).
1
I am personally unsure about how/why to use hypergraphs; for the problems I've been modelling simple graphs seem sufficient. Note you can model a relationship between n nodes in graphs by adding a "hub" node that represents the relationship, call it R, and linking every n with R.
1