Many real-world data-sets can naturally be framed as graphs. For example, on online platforms such as social networks, users can be represented as nodes, and follows or likes can be represented as edges.
However, when building models on data from these domains, people often simplify the problem by ignoring the underlying graph structure. In doing so, machine learning practitioners ignore useful information that would help contextualize an entity (e.g. a user) in the context of the broader network they are a part of.
What are graphs?
Graphs are data structures that encode relationships between pairs of entities. Entities in the graph are referred to as nodes.
