【ML | Graph Data】Node Classifica

Homophily: Homophily is a concept in sociology describing the tendency of individuals to associate and bond with similar others.

Define a set of node features (e.g. Node attributes, aggregate over links (collaboration) and boolean function based on graph neighborhood.)

For each node, construct the training example (node features + label) - Use a standard machine learning approach to learn a classifier from the training examples.

For nodes: calculate feature vector X(u), and predict Y(u) -

Essentially inductive, but could also be used in a transductive situation

No node features / Transductive classification exploiting homophily

For unlabeled nodes compute probability distribution over class labels

Iterative procedure: unlabeled nodes calculate proportions of neighbors labels.