What are networks (and why should you care)?

What are networks?

  • Networks are a mathematical way of representing a set of objects and the relationships among them.
  • These "objects" are termed nodes or vertices.
  • These relationships are termed edges or, less often, links.
  • Networks are also called graphs.

Example networks / applications

Connectome

center

More connectome

center

Enron anomaly prediction

center

Organizational communication

Faculty hiring and other hierarchies

Pandemic spread


Protein-protein interaction

center

What are NOT networks? - Hypergraphs

  • Networks represent dyadic relationships: interactions between two things.
    • Example: an email from me to you
  • Polyadic relationships (interactions between more than two things) are common.
    • Example: an email from me to you AND someone else
  • We could ignore this:
    • Make an edge from me to you.
    • Make another, separate edge from me to someone else.
  • Hypergraphs are a mathematical way of representing general polyadic relationships.

What are NOT networks? - Multigraphs

  • Graphs (strictly speaking) usually have at most one edge between node ii and node jj.
  • There may be multiple relationships between two nodes in data that we want to model.
    • Example: An email from me to you, and a phone call from me to you.
  • Sometimes we can compress this information into at most one edge, and still use a graph.
    • Example: Create an edge if there was an email OR a phone call.
  • Multigraphs allow for more than one edge from node ii to node jj.

Every time we represent something in the real world with a network, we're making a modeling choice

What can we do with networks? (i.e. What is this class about?)

Class Calendar

Nodes: email accounts, edges: some # of emails