Telecon 4.12.2015

Simple Filter Query

Daniele's Proposal https://dl.dropboxusercontent.com/u/14442682/temperature-queries.pdf

Captures execution policy explicitly by using query
Output timestamp needs to be considered
- Could provenance trace be used to capture input timestamps and use evaluation time for the output graph
What should be the output on the stream when there is no matching data value in the input stream?

Answer for time point 13 assumes a computation is made every minute and consumed values are ignored.

Tara's Proposal http://lists.w3.org/Archives/Public/public-rsp/2015Dec/0000.html

Focused on the application of the SPARQL query to the RDF dataset
Does not require a window operator

Assume that items are consumed as they arrive on the stream

Comparison of approaches

For simple filter query do we need a window?
Do we need to preserve the SPARQL semantics, particularly when we consider window to stream operators?
- Tara's approach has an implicit windowing operation included within it.
How do we access the timestamps of the items in the stream to be able to do comparisons on these?

Concerns about the Group

Community group has been running for a long time now
No clear consensus has yet come out of the group
We need to consider how we can speed the processes up
We don't have a committed core and keep having new use cases being put forward
Need to have issues being addressed and agreed at each meeting
Need to generate a report
Need to find a communication platform which all people can use (constant problems with voice with Adobe)

Next Meeting

18 December

Discuss two approaches with respect to the use case using Tara's data stream, can we pose the two different queries: average over the cities at each time point and the average over the last two time points?

We should also consider the case where we have a simple join over two streams.