Though I was not able to attend the 1st International Workshop on Stream Reasoning during ESWC2009 (as I was co-chairing our SPOT workshop) I did enjoy the workshop wrap-up and summary on the last day. Each workshop organizer had ten minutes to talk about how the workshop went, what the upcoming challenges are, etc.
Now, I was sitting in this session and thinking about what from the linked data perspective there is potentially in it. I know, it’s a very limited thing to do: for a hammer, everything looks like a nail, right? ;)
However, let’s for a moment put on our linked data glasses and see what we can find. I’ve earlier argued that microblogging is sort of linked data in disguise. Many more of these ‘micro-content’ things on the Web come to mind: news feeds in Atom or RSS, chat logs, one could even understand mailing lists as sort of slowly progressing, rather heavyweight streaming data sources, etc.
Looking at all these sources, they have a couple of things in common one needs to address in order to apply/use linked data on/with it:
- addressing the temporal dimension: one needs to keep track of when something happened
- addressing the order: it can be quite important to know in which order something happened
- addressing the provenance: one has to keep track where something came from
There are for sure other issues I’ve not been thinking so far, but from the short list above one can already guess that we’ll have to invest some more resources and energy to address the streaming aspect from a linked data perspective.
Let me know your favorite solution to the issues above (some known proposals/work are beneath others: named graphs – hey, this should be in RDF core by now, right? – as well as Olaf’s great work on provenance).