Optimal Display Adaptation of Iconic Document Visualizations via BFOS-Style Tree Pruning

This paper introduces a new visual representation of a document or group of documents, a dynamic document icon, or Dydocon. Its representation is symbolic like an icon, but changes depending on document content. A Dydocon can be used for multiple documents and thus is useful for presentation of clustered search results. Because a large number of clusters may be returned by a search it is desirable to optimally select clusters given available display space. We cluster documents into a visual taxonomy described by a tree of Dydocons, define tree functionals for icon size and distortion of a cluster, and perform display-size adaptation through optimization using a BFOS-style algorithm.