Interactive Steering of Hierarchical Clustering

Hierarchical clustering is an important technique to organize big data for exploratory analysis. However, existing one-size-fits-all hierarchical methods often fail meet the diverse needs of different users. To address this challenge, we present interactive steering method visually supervise constrained by utilizing both public knowledge (e.g., Wikipedia) and private from The novelty our approach includes 1) automatically constructing constraints using (knowledge-driven) intrinsic distribution (data-driven), 2) enabling through a visual interface (user-driven). Our first maps each item most relevant items in base. An initial constraint tree then extracted ant colony optimization algorithm. algorithm balances width depth covers with high confidence. Given tree, are hierarchically clustered evolutionary Bayesian rose tree. clearly convey results, uncertainty-aware visualization has been developed enable users quickly locate uncertain sub-hierarchies interactively improve them. quantitative evaluation case study demonstrate that proposed facilitates building customized trees efficient effective manner.