Set vs Graph-based visualisations of multiple trees

24
A Comparison of Set-based and Graph-based Visualisations of Overlapping Classification Hierarchies Martin Graham, Jessie Kennedy & Chris Hand Napier University, Edinburgh

Transcript of Set vs Graph-based visualisations of multiple trees

Page 1: Set vs Graph-based visualisations of multiple trees

A Comparison of Set-based and Graph-based Visualisations of Overlapping Classification

Hierarchies

� Martin Graham, Jessie Kennedy & Chris Hand

� Napier University, Edinburgh

Page 2: Set vs Graph-based visualisations of multiple trees

Overview

� Introduction� Problem Domain� Similar Approaches� Prototype 1 - Graph� Prototype 2 - Set (+ example)� Testing� Conclusion

Page 3: Set vs Graph-based visualisations of multiple trees

Introduction

� Hierarchical information structures are a major topic within Information Visualisation (IV)

� Cone Trees - Robertson, Mackinlay & Card� Tree Maps - Johnson & Shneiderman

� Multiple hierarchies relatively unexplored� Relations (intra- and inter- tree)

Page 4: Set vs Graph-based visualisations of multiple trees

Taxonomy

� Re-classifications of plants� A B

� C D

Page 5: Set vs Graph-based visualisations of multiple trees

Tasks for Multiple Taxonomies

� Taxonomist Requirements� Track a particular genus’s siblings and

parents� Track a higher level (internal) node’s children� Compare depths� Compare global structures

Page 6: Set vs Graph-based visualisations of multiple trees

Previous techniques

� Animation� Strong pre-attentive cues� Suited for gradual, not abrupt, changes� Shows change between only 2 states at any

one point� Small Multiples (cf Tufte)

� Display all structures� ‘N’ small multiples = 1/’N’th screen area each

Page 7: Set vs Graph-based visualisations of multiple trees

Previous Techniques

� Wittenburg et al’s Group Asynchronous Browsing on the WWW� Combines multiple TreeMaps on top of a

MultiTree structure� Shows correlations (cross-references) using

colour� Static colouring only� Can show top-level correlations but not for

individual parts of hierarchies

Page 8: Set vs Graph-based visualisations of multiple trees

Graph-based prototype

� Combine multiple hierarchies into one graph structure

� Visualised using spring-mass metaphor� Different coloured links & glyphs indicate

relationships in different hierarchies� Interaction via panning/zooming and

filtering of a particular node’s relations

Page 9: Set vs Graph-based visualisations of multiple trees
Page 10: Set vs Graph-based visualisations of multiple trees
Page 11: Set vs Graph-based visualisations of multiple trees

Graph-based Prototype

� Advantages� Integrate multiple hierarchies spatially and

temporally� Drawbacks

� Above! Double-edged sword� Layout issues� Computational complexity

Page 12: Set vs Graph-based visualisations of multiple trees

Set-based prototype

� Sets reflect mental model of taxonomic reasoning

� Differentiate between categories and the objects we are organising

� Use colour via linking to indicate correlations

� Linking applied over a set of hierarchies

Page 13: Set vs Graph-based visualisations of multiple trees
Page 14: Set vs Graph-based visualisations of multiple trees

Set-based Prototype

� Advantage of� Reflects taxonomist’s mental model

� This is at the price of� Lose a level of detail (unnamed leaves and

only one relation set shown at a time)� Have to accept restrictions of small multiples

� However...� Clearer visualisation

Page 15: Set vs Graph-based visualisations of multiple trees
Page 16: Set vs Graph-based visualisations of multiple trees
Page 17: Set vs Graph-based visualisations of multiple trees

Prototype Effectiveness

� Metrics v Empirical Testing� Brath’s metrics� Suitable for static visualisations� Interactive visualisations cause problems� Zooming and filtering reduce on-screen data

density (but make for a more relevant picture)

Page 18: Set vs Graph-based visualisations of multiple trees

Prototype Effectiveness

� Informal User Testing (cf Nielsen)� A handful of typical would-be users� Used early in development� Task-based observations generate solutions

(Monk & Wright)

Page 19: Set vs Graph-based visualisations of multiple trees

Prototype Effectiveness

� Observations� Overlapping (occluding) nodes caused

confusion in graph prototype� Non-leaf (category) nodes in set based

prototype indicated by highlighting their descendent leaf nodes, confusing users as to extent of the non-leaf node

Page 20: Set vs Graph-based visualisations of multiple trees

Prototype Effectiveness

� Solutions� Ensure that highlighted nodes always

displayed on top of unselected nodes� Allow user to fine-tune positioning.

� Non-leaf nodes and leaf nodes to be highlighted in similar fashion

Page 21: Set vs Graph-based visualisations of multiple trees

Prototype Effectiveness

� Outcome� Definite preference for set-based visualisation

prototype� Graph-based visualisation prototype too

cluttered, even with zooming/filtering mechanisms

� More tasks envisaged

Page 22: Set vs Graph-based visualisations of multiple trees
Page 23: Set vs Graph-based visualisations of multiple trees

Conclusions

� Constructed 2 prototypes for visualising multiple hierarchies and their correlations

� Users stated both prototypes enabled them to see information that is not possible with their current system

� Future work� Test prototypes on other information sets

Page 24: Set vs Graph-based visualisations of multiple trees

Acknowledgements

� Royal Botanic Garden Edinburgh� EPSRC

� http://www.dcs.napier.ac.uk/~marting