This article illustrates the possibilities arising from applications which store user data, or which build databases on SAFE with LinkedData (a semantic data format allowing you to run semantic queries).
It’s actually about querying multiple data sets (SPARQL endpoints) simultaneously, in order to produce results which are mashups of data from multiple databases and, wait for it… most of this happens in the client (browser) making this a very interesting possibility for SAFE apps.
To achieve exactly what this article describes (it includes a live demo too so you can try it) we need to figure out a way to handle the single triple based query per data set, used in their examples. At this point I don’t know how feasible that is.
I’ve not heard of this before - thanks for sharing!
This sounds like something we should integrate with Safe CMS to generate the REF data automatically. This sort of thing would surely contribute towards good distributed search engine support too.
Interesting. This doesn’t really solve the issues of scale and downtime on the web though, just alleviates them a tiny bit. SAFE does however solve these issues, so if this could be made to work on SAFE it could be quite useful.
There’s an example query linked there which is to find subjects with object “restaurant” and predicate “type”. In that case there could be a mutable data that’s looked up with “type restaurant” and that contains the list of all restaurants. Similar indexes could be made for the other combinations. Maybe a mutable data could have a name that is “predicate:type” and then under that keys like “restaurant”, “cafe” etc. and then again each of those having a list of subjects.