Visualization of workflow results

Typical workflows will contain processes that generate graphs showing intermediate and final results. These are often useful to spot inconsistencies in the data, troubleshoot or simply communicate the results effectively.

For research purposes, graphics do not need extensive customization. As long as axes are labelled, markers and lines are identified by a legend and a title can be assigned, most users will be happy with the results. PAVICS should provide a suite of typical graphing processes that can be embedded within a workflow or run in a separate workflow once the results are available. So for example, researchers might want to define a compute workflow (bundling all computation processes), and an analyze workflow (bundling statistical syntheses, tables and figure creation).

For publication purposes, graphics are often customized to achieve a more polished finish (custom colours, line widths, axes limits, etc.), especially for maps. As long as the data is easily accessible to recreate a graph, we should expect that users will be responsible for creating those publication-ready graphs. Having the graphic generation code easily available would help in the process, as would data outputs in portable formats (e.g. JSON). So, for example, in the analyze workflow, users could call the visualization process to return the figure and the code generating the figure.

Dynamic sandbox visualization

Once results are available, researchers often want to play with the data and explore different ways to graphing it. This implies an environment where users can select the programming language they are the most comfortable with (e.g. Matlab/Octave, Python, R). This also likely means being able to quickly generate widgets allowing for more dynamic ways of interacting with the data (SVG, HTML5).

Custom frontend

For popular and more public results, we will want to display results that have been produced by a workflow on a dedicated web site.