Tip:
Highlight text to annotate it
X
DiffAni is a hybrid visualization for dynamic graphs.
It allows a user to use different visualization techniques
over different time slices. The dynamic graph
is displayed as a sequence of tiles. In this initial view,
each tile shows one time slice of the data.
These are called small multiple
tiles. Our DiffAni prototype also supports
two other kinds of tiles called animation tiles
and difference tiles. The user can convert some of
these small multiple tiles to other kinds of tiles
by dragging with the mouse to select tiles
and then dragging a pig-tail gesture, causing a radial menu
to pop up. The user may convert the selected tiles
to any of the three kinds of tiles. If they choose
animation tiles, a smooth transition shows
the tiles converting to a single animation
tile. Animation tiles are shown with a dashed
border. We see that the animation tile
covers time slices 4,
5, and 6. The user can select
other tiles and convert them to a different style
which highlights differences between two time slices,
in this case between time slices 2 and
3. Also notice that there is a dashed vertical line on the left.
That is the time cursor showing our current position in time.
When the mouse cursor is placed over a node, that node and its neighbors are
highlighted in the current tile. In the case of difference maps,
red color is used to indicate removal of a node
or an edge, while green denotes a new element.
If the Alt key is pressed at the same time,
it also highlights the nodes in the full graph history.
Small multiples represent a
single time slice, while diff and animation can represent several time slices.
The user can browse the graph history by dragging his mouse back and forth
in a fast or slow manner. This scrubbing technique
works the same way for all tiles. In the case of an animated
tile, the user can move in time by dragging the mouse
inside a tile.
We saw earlier that a user can
compress or combine several tiles together. However, tiles can also be
decompressed
to small multiples.
Our prototype allows to flexibly map representations
to transitions of a dynamic network. Here,
we are compressing time slices 2 and 3 to a difference map,
and time slices 6 and 7 are mapped
to animation.
Then we convert
slices 4 and 5 to a difference map.
To explore alternative hybrid visualizations,
we could also decompress tiles to small multiples
as a first step.
In a second step, we could also convert
some of these to other representations
using the pig-tail interaction.