Characterizing measures for the assessment of cluster analysis and community detection
Résumé
The problem of comparing two partitions of the same set occurs in a number of situations, the most widespread being probably the assessment of cluster anal- ysis and community detection results. In these contexts, one has computed the clusters of a dataset, or the community structure of a network. This result takes the form of a partition of the set of data points or set of nodes, respectively. One then wants to compare this estimation with some ground-truth also taking the form of a partition. Alternatively, one has computed several such estimations, for instance using several algorithms, and wants to compare them to each other.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |