Section: Overall Objectives
- Hybrid Matrix Representations
Nathalie Henry and Jean-Daniel Fekete have published two important articles on network visualization using hybrid representations of matrices and node-link diagrams: MatLink [Oops!] and NodeTrix [Oops!] . Finding better representations for large and dense networks is important to support exploration tasks for domains such as social network analysis and bioinformatics. Our representations have very good properties and were welcome by the HCI and InfoVis community. MatLink received the best paper award at the Interact 2007 conference (Brian Shackel Award). See section 6.1 for details.
- Multi-scale Navigation in Huge Networks
AVIZ has reached a new milestone in multi-scale visualization and navigation of large networks with ZAME, the Zoomable Adjacency Matrix Explorer [Oops!] that can reorder and visualize network with millions of vertices and tens of millions of edges. We intend to use it to explore the Wikipedia networks (article to article, article to author, author to author, etc.) and large protein networks in bioinformatics (see section 6.2 ).
- Visualizing Wikipedia for Occasional Users
Stéphane Huot and Jean-Daniel Fekete have designed new visualizations called WikipediaViz to improve the trust and transparency of Wikipedia articles.
- SVM for Very Large Datasets
Thanh-Nghi Do has substantially increased the performance of SVM classification algorithms [Oops!] . His method is applicable to very large dataset in the number of individuals and dimensions (see section 6.4 ).