Can animation support the visualisation of dynamic graphs?by Daniel Archambault, Helen C. Purchase

Information Sciences


An Account of the State Prison, or Penitentiary House in New-York (Concluded)

One of the Inspectors of the Prison

The status of the otter (L. lutra L.) in Britain in 1977


Principles, concepts, and practices in prosthodontics

The Academy of Denture Prosthetics

Christians and atomic war

The British Council of Churches


Available online xxxx


Graph drawing

Dynamic graphs states in separate windows. The user scans these windows to see how the data evolves. ith th animation nimation t the use of a slider and a play/pause button. Smooth linear interpolations transition node positions between timeslic nodes and edges are faded in/faded out if they are inserted/removed from the data. A small multiples presentation graph depicts each timeslice of the data in its own window [55]. The windows are arranged in chronological orde 0020-0255/ 2015 Elsevier Inc. All rights reserved. ⇑ Corresponding author at: Department of Computer Science, Faraday Tower, Swansea University, Singleton Park, Swansea SA2 8PP, United Kingdom. Tel.: +44 1792 295393; fax: +44 1792 295708.

E-mail addresses: (D. Archambault), (H.C. Purchase).

Information Sciences xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Information Sciences journal homepage: www.elsevier .com/locate / inssmall multiples are the two most common, basic methods for the visualisation of dynamic graphs. An dynamic graph presents the data like an interactive movie whereby the user has complete control of the aPlease cite this article in press as: D. Archambault, H.C. Purchase, Can animation support the visualisation of dynamic graphs?, Info (2015), the hrough es and of the r like aedges can be added and removed. Dynamic graphs are usually represented using a series of timeslices or time steps whereby snapshots of the graph are taken over given time intervals. For graphs with this representation, dynamic graph drawing algorithms assign coordinates to the nodes in the timeslices in order to depict the structural evolution of the graph over time. Dynamic graphs exist in many application areas including: software engineering [17], computer networks [12], distributed systems [47], sociology and social networks [57,14], systems biology [9], financial networks [58] and many others. Understanding the perceptual factors related to visualising dynamic graphs is important for these fields.

An important factor to consider when visualising a dynamic graph is how the information is presented. Animation andMental map

Drawing stability


Small multiples 1. Introduction

Dynamic graph drawing deals wIn a recent experiment, drawing stability (known more widely as the ‘‘mental map’’) was shown to help users follow specific nodes or long paths in dynamically evolving data.

However, no significant difference between animation and small multiples presentations was found. In this paper, we look at data where the nodes in the graph have low drawing stability and analyse it with new error metrics: measuring how close the given answer is from the correct answer on a continuous scale. We find evidence that when the stability of the drawing is low and important nodes in the task cannot be highlighted throughout the time series, animation can improve task performance when compared to the use of small multiples.  2015 Elsevier Inc. All rights reserved. e depiction of graphs that evolve over time in terms of their structure. Nodes andCan animation support the visualisation of dynamic graphs?

Daniel Archambault a,⇑, Helen C. Purchase b a Swansea University, Department of Computer Science, United Kingdom bUniversity of Glasgow, School of Computer Science, United Kingdom a r t i c l e i n f o

Article history:

Received 1 November 2014

Received in revised form 2 March 2015

Accepted 9 April 2015 a b s t r a c t

Animation and small multiples are methods for visualising dynamically evolving graphs.

Animations present an interactive movie of the data where positions of nodes are smoothly interpolated as the graph evolves. Nodes fade in/out as they are added/removed from the data set. Small multiples presents the data like a comic book with the graph at variousrm. Sci. 2 D. Archambault, H.C. Purchase / Information Sciences xxx (2015) xxx–xxxcomic book to depict the evolving data. This presentation method has been shown to be more effective than an animation for dynamic graphs on a variety of tasks [6,25]. More specifically, small multiples is faster with no statistically significant difference in terms of error rate when linked highlighting between timeslices is used.

Animations have the advantage that the full screen can be devoted to a single timeslice at a given time. However, interaction is required and the user must rely on memory to understand how the graph evolves. Small multiples has the advantage that all timeslices are visible on the screen simultaneously and the user does not need to rely on memory. However, the number of pixels devoted to each timeslice is much less and could potentially make the graph more difficult to read.

In dynamic graph drawing, there is always a compromise between the drawing quality of the graph in each timeslice and the stability of the node positions in the drawing as the graph evolves through time. Preserving the mental map [23,41], or drawing stability, is a property of the dynamic graph that does not allow nodes to move too far in the plane between consecutive timeslices when a change is made to the graph. Thus, stable parts of the graph structure remain in similar positions in the plane during graph evolution. Recent experiments have shown that drawing stability can help with tasks that involve determining the insertion order of specific nodes [29], revisitation to specific nodes in a graph at later times [3], and following long paths through the data [3].

For tasks where drawing stability is important and the positions of nodes remain relatively stable throughout graph evolution, animation can provide little benefit as the identity of each node is disambiguated through its position. However, if the nodes of the drawing move substantially as the graph evolves, animated transitions might provide a larger benefit when compared to small multiples. Situations where the structure of the graph changes substantially due to node insertions/deletions can be the cause of such movement. In many areas of the graph visualisation and information visualisation literature [10,46,8], animated transitions have been shown to be helpful when there are significant changes to the data and/or how it is arranged on the screen.