One of the outcomes of our recent Advisory Group Meeting was that I needed to do some work on the paradata “recipes” we want for SPAWS. These recipes are the ways we create the JSON activity stream data for particular paradata “verbs”.
So far I have recipes for:
- Rating: A single rating for a widget
- Review: A single review or comment on a widget
- Ratings: Aggregated average ratings for a widget
- Stats: Aggregated statistics for a widget
I originally had a review comprised of both comment and rating, like Amazon, however looking at Edukapp, ITEC and ROLE widget stores they seem to keep comments and ratings separate so I’ve factored them out.
Also from the advisory group was the idea of including contextual information so you can obtain information about the paradata in context; so each JSON record can also provide the URL of the page in the store for the widget details, as this is where the originating information will be displayed. This can be useful for recommenders, for example, by looking at what else is on the page using a scraper or microdata.
For stats (downloads, embeds, likes etc) I’ve assumed a model of periodically publishing the latest aggregate total rather than monthly or weekly summaries as this would seem more robust. Perhaps the context URL for stats could point to a dedicated stats page with breakdowns for more detailed analysis; however I think the most common use is going to be to provide cross-store totals.
Once we’ve tested these recipes out, we could then add them to the Learning Registry Paradata Cookbook.