Developing quality measures for clustering constraints and examining where existing approaches fall short.
Constrained clustering
Conceptual illustration. Ongoing research examines criteria for assessing the quality and potential benefit of clustering constraints. The diagram poses the research question; it does not claim an established answer. Open full size ↗
Which constraints actually help a clustering algorithm? This recurring question still lacks a sufficiently general and accurate answer.
My current research examines limitations in existing methods and develops ways to assess the benefit of a constraint. The aim is to reason more reliably about constraint quality, rather than assuming that every supplied constraint improves a clustering.
