If you click on the Datastore link in the console's navigation panel, you'll see this:
The first two columns make good sense: we have one DatastoreRegexp instance, with name dataset and value set-[0-9]. The last two columns aren't self-explanatory, but to find out what they are and how they work, double-click the dataset row. You'll see the following dialog box:
It looks like you can enter text into these boxes, and indeed you can:
If you now press Save, here's what you see back on the main panel:
If you were now to run the sample pipeline, you would notice something interesting: Ziggy only creates one task for each node, and that task is the set-1 task! What you have done by adding set-1 as an include regexp is you've added a condition to the dataset DatastoreRegexp: when it sweeps through the directories to generate units of work, the dataset level directories need to match the dataset value but also match its include regexp.
The exclude regexp, by symmetry, does the opposite: only dataset level directories that do not match this regular expression can be included. Rather than setting the include to set-1, we could have left the include blank and set the exclude to set-[2-9].
Going back to the ludicrous example from the Instances Panel article, we can now see how we would go about limiting the pipeline to running only tasks where guitar equals reeves and album equals either outside or stardust. We would go to the regular expressions panel and set the guitar DatastoreRegexp include value to reeves; we would then set the album include regexp to outside|stardust.