HP Labs Technical Reports
Click here for full text:
Visual Mining Large Web-based Hyperbolic Space Using Hidden Links
Hao, Ming; Hsu, Meichun; Dayal, Umesh; Krug, Adrian
Keyword(s): visualization; hyperbolic space; data mining
Abstract: Many real-world KDD (Knowledge Discovery & Data Mining) applications involve navigation of large hierarchies, for example, internet resources, hot topics on the web, organization charts. Quite often users feel lost, confused, and overwhelmed with many complex displays. This paper discuss a new dynamic "hidden link" technique for linking a large highly connected graph in a simple hyperbolic space without cluttering the display. Only the primary directed non- cyclic links are shown. All other types of links (non- tree/cross-link) are hidden in the property of each node. These hidden links only appear when the user navigates/clicks on the node. "Hidden Link" allows a user to freely focus on the hierarchy of interest.
Back to Index