A Survey of Clustering with Instance Level Constraints

Clustering with constraints is an important recent development in the clustering literature. The addition of constraints allows users to incorporate domain expertise into the clustering process by explicitly specifying what are desirable properties in a clustering solution. This is particularly useful for applications in domains where considerable domain expertise already exists. In this first extensive survey of the field, we discuss the uses, benefits and importantly the problems associated with using constraints. We cover approaches that make use of constraints for partitional and hierarchical algorithms to both enforce the constraints or to learn a distance function from the constraints.