MultiLayerMatrix: Visualizing Large Taxonomic Datasets

Dang, Tuan Nhon; Cui, Hong; Forbes, Angus G.

MultiLayerMatrix: Visualizing Large Taxonomic Datasets

Files

055-059.pdf (4.35 MB)

0115-file1.mp4 (40.12 MB)

Date

2016

Authors

Dang, Tuan Nhon
Cui, Hong
Forbes, Angus G.

Publisher

The Eurographics Association

Abstract

Adjacency matrices can be a useful way to visualize dense networks. However, they do not scale well as the network size increases due to limited screen space, especially when the number of rows and columns exceeds the pixel height and width of the screen. We introduce a new scalable technique, MultiLayerMatrix, to visualize very large matrices by breaking them into multiple layers. In our technique, the top layer shows the relationships between different groups of clustered data while each sub-layer shows the relationships between nodes in each group as needed. This process can be applied iteratively to create multiple sub-layers for very large datasets. We illustrate the usefulness of MultiLayerMatrix by applying it to a network representing similarity measures between 2,048 characters in the Asteraceae taxonomy, a rich dataset that describes characteristics of species of flowering plants.We also discuss the scalability of our technique by investigating its effectiveness on a large synthetic dataset with 20,000 columns by 20,000 rows.

        @inproceedings{10.2312:eurova.20161125
,
booktitle = {EuroVis Workshop on Visual Analytics (EuroVA)
},
editor = {Natalia Andrienko and Michael Sedlmair
},
title = {{MultiLayerMatrix: Visualizing Large Taxonomic Datasets
}},
author = {Dang, Tuan Nhon and 
Cui, Hong and 
Forbes, Angus G.
},
year = {2016
},
publisher = {The Eurographics Association
},
ISSN = {-
},
ISBN = {978-3-03868-016-1
},
DOI = {10.2312/eurova.20161125
}
}