US 7,483,907 B2
Processing data-stream join aggregates using skimmed sketches
Sumit Ganguly, Bhopal (India); Minos Garofalakis, Morristown, N.J. (US); and Rajeev Rastogi, New Providence, N.J. (US)
Assigned to Alcatel-Lucent USA Inc., Murray Hill, N.J. (US)
Filed on Dec. 29, 2004, as Appl. No. 11/25,578.
Prior Publication US 2006/0143170 A1, Jun. 29, 2006
This patent is subject to a terminal disclaimer.
Int. Cl. G06F 17/30 (2006.01)
U.S. Cl. 707—101  [707/2; 707/6; 707/100] 14 Claims
OG exemplary drawing
 
1. A method of estimating a join aggregate over one or more data-streams, the method comprising the steps of:
maintaining one or more atomic sketches over said one or more data-steams;
skimming dense frequencies from said atomic sketches, thereby creating a dense frequency vector and an atomic skimmed-sketch for each of said one or more data-streams, and;
generating an atomic estimate of said join aggregate using said atomic skimmed-sketches and said dense frequency vectors and storing said atomic estimate in a computer memory as an approximation of a size of said join aggregate.