Social Network Quantification


Quantification is a task strictly related to Classification, yet it aims to estimate the class distribution on new unseen instances and it focuses on capturing distribution drifts rather than classifying correctly individual instances.
Quantification & Social Networks
Problem Given a partially node-labeled social graph infer the label distribution of the unlabeled entities.
  1. Community based Quantification
    Identify communities on the full graph and define some strategies to assign labels to unlabeled nodes.
  2. Ego-network based Quantification
    Exploit only the Ego-network of unlabeled nodes to identify the correct label.

Komentar

Postingan populer dari blog ini

Kevin Bacon's Ego Network

Artificial Intelligence