Efficient Join Processing over Uncertain Data - By Reynold Cheng, et all.

1
Efficient Join Processing over Uncertain Data -
By Reynold Cheng, et all.

• Presented By
• Lydia Usha

2
Main Idea
The main key point addressed in this paper is
extending traditional join techniques to DBs with
uncertain attributes. Improving the efficiency of
join based algorithms to address some of the
complexities faced by DB management in handling
joins over uncertain data. Efficient pruning
techniques involving both uncertainty interval
and uncertainty pdf.
3
Importance of the problem

• Often spatial DB applications have to deal with
  uncertain data. For example
• GPS data which calculates nearest neighbour
• Sensor data from realtime applications
• Scientific data corresponding to weather.

4
Key concepts
Addresses the semantic complexities of uncertain
data Using probabilistic threshold joins. Define
uncertainty comparison operators probabilistic
join queries. Proposes novel techniques like page
and index level joins which incorporates
efficient pruning techniques and evaluating the
performance of those algorithmsover uncertain
data.
5
(No Transcript)
6
Probabilistic uncertainty model
Uncertainty interval uncertainty pdf.
7
Item Level Join
8
Page level pruning
9
Page level pruning
10
Index level Join
Used to improve IO throughput When combined with
node level pruning technique, this improves
performance as well as IO throughput. In this
pages are organized in an ordered tree structure,
allowing one to use MBR which tightly encloses
the interval within the subtree.
11
Experimental Result.