We describe Hokusai, a real time system which is able to capture frequency information for streams of arbitrary sequences of symbols. The algorithm uses the CountMin sketch as its basis and exploits the fact that sketching is linear. It provides real time statistics of arbitrary events, e.g. streams of queries as a function of time. We use a factorizing approximation to provide point estimates at arbitrary (time, item) combinations. Queries can be answered in constant time.

Author comments: Appears in Proceedings of the Twenty-Eighth Conference on Uncertainty in Artificial Intelligence (UAI2012)Report no: UAI-P-2012-PG-594-603

Proxy: auai

http://arxiv.org/abs/1210.4891

http://arxiv.org/pdf/1210.4891.pdf

## 0 comments