Constraint-based clustering in large databases

摘要: Constrained clustering--finding clusters that satisfy user-specified constraints--is highly desirable in many applications. In this paper, we introduce the constrained clustering problem and show traditional algorithms (e.g., k-means) cannot handle it. A scalable constraint-clustering algorithm is developed study which starts by finding an initial solution satisfies constraints then refines performing confined object movements under constraints. Our consists of two phases: pivot movement deadlock resolution. For both phases, optimal NP-hard. We propose several heuristics how our can scale up for large data sets using heuristic micro-cluster sharing. By experiments, effectiveness efficiency heuristics.