Abstract
In the last week, we have an interesting discussion in the basic
architecture of the stream model. In this talk,
I will focus on two fundamental issues:
(1) Memory boundness: Can we decide whether a continuous query
can be evaluated using a bounded amount of memory for all
possible instances of data streams?
(2) Data Sketching: Given that approximate answers are allowed, we
discuss how the technique of "data sketching" can be used to
answer a "self-join" query efficiently with an upper error bound.