This document helps you work through an example of visualizing time series network data using animation in KP 4.
Start running KP by double-clicking the file 'kp.bat'. You will need java on your computer, preferably version 1.5, though earlier versions will work. Expand the window to fill the screen.
First, load a series of snapshots using the file!open command. In the file dialog that this command opens, go to the newcomb directory included with the archive and load the file 'series.kps'. This names a set of files, one per line, that define graphs in the series. Each graph is in a simplified ucinet dl-style format that allows spaces between the matrix entries, so link values can go above 9.
There is currently a bug so that you have to resize the screen to see anything after the slider has been added. I suggest expanding the screen to fill the whole desktop. Then drag the slider down to occupy the lower quarter of the screen. You will see one graph, laid out randomly, in the main window and the phrase 'graph 1 of 15' in the title bar. The slider will have 16 positions.
Lay out the whole series of graphs by selecting 'Series!Anneal series' from the menubar at the top. Nothing will be shown until all the graphs are laid out, except for the title bar which will show the graph currently being worked on. When they are all laid out, the last graph in the series will be shown. You can use the left and right arrow keys to move through the series of graphs. They wrap around, so the right arrow key takes you from the last graph to the first one. When you click on the slider itself, it grabs the keyboard input and the left and right keys just increment the slider, which moves between the graphs very slowly. If you have touched the slider, use it from now on to move between the graphs. As you drag it, the graph nearest the time point (times 100) will be shown - e.g. the first graph is shown when the slider is at 0, the second when it gets near 100, and so on.
To see a smooth morph, make sure the series is showing the first graph, and select 'Series!morph series'. Morphing always goes from the first graph shown until the end of the series, so nothing will happen if the graph shown is the last one in the series.
If the graph moves too quickly or too slowly, you can adjust the number of frames in each segment through the 'Series' menu.
It is easier to follow the nodes if they are colored. To make each node have a different color, select 'Display!Style!set node fill color', then choose 'name', so the color is computed based on the name. You can also make the colors depend on the indegree of the node, for instance. Select 'Display!Style!set node fill color', then choose 'indegree'. The numbers are automatically translated to colours so that 0 is dark blue, the maximum value is red and the average value is white. All the graphs in the series have been changed, so if you morph again you will see the colors change.
You can show any node property graphed over time. First, clear the timeline by selecting 'Timeline!clear'. Now, right click over the node called 'Ralph' and select 'Graph node feature over the series', then select 'indegree'. The same function that is being used to colour the nodes is now graphed over time for Ralph in the lower window. Several nodes can be graphed at the same time, for example, David has an interesting interplay with Ralph.
This feature is not enabled in the current version
Many of the node and network features available on the menu are defined in Carter Butts' SNA package written in R. KrackPlot passes the network to R while it is running and reads back the values for these features. To do this, you need to have R running on your system with Carter's SNA package installed. To install the package on windows, run R with the GUI, and select 'Packages!install package(s) from CRAN..', the select 'sna'. Next, you need the RServe package from http://stats.math.uni-augsburg.de/Rserve/ . Follow the instructions there - I have only tried this on windows. You will need to have an instance of the R server running when you try to access an R command from KrackPlot. In windows, do this by running the command 'Rcmd.exe Rserve.exe' from the command prompt. KrackPlot will load SNA when it calls a function, assuming that you have installed it. If you have got this far, you can test it out by graphing the nodes' betweenness centrality rather than the indegree.