Docsity
Docsity

Prepare for your exams
Prepare for your exams

Study with the several resources on Docsity


Earn points to download
Earn points to download

Earn points by helping other students or get them with a premium plan


Guidelines and tips
Guidelines and tips

Spatial Joins - Project Proposal - Database Management Systems | CS 236, Study Guides, Projects, Research of Computer Science

Material Type: Project; Class: DATA BASE MANAGEMENT SYSTEMS; Subject: Computer Science; University: University of California-Riverside; Term: Unknown 1997;

Typology: Study Guides, Projects, Research

2009/2010

Uploaded on 03/28/2010

koofers-user-uv1
koofers-user-uv1 🇺🇸

10 documents

1 / 9

Toggle sidebar

This page cannot be seen from the preview

Don't miss anything!

bg1
1
1
Spatial Joins
2
Papers to Present
•“Efficient Processing of Spatial Joins using
R-trees”, T. Brinkhoff, H-P Kriegel and B.
Seeger, Proc. SIGMOD, 1993.
•“Spatial Joins using Seeded Trees”, M-L Lo
and C. V. Ravishankar, Proc. SIGMOD,
1994.
•“Spatial Hash-Joins”, M-L Lo and C. V.
Ravishankar, Proc. SIGMOD, 1996.
•“Size Separation Spatial Joins”, N. Koudas
and K. C. Sevcik, Proc. SIGMOD, 1997.
pf3
pf4
pf5
pf8
pf9

Partial preview of the text

Download Spatial Joins - Project Proposal - Database Management Systems | CS 236 and more Study Guides, Projects, Research Computer Science in PDF only on Docsity!

1

Spatial Joins

Papers to Present

  • Efficient Processing of Spatial Joins using R-trees ”, T. Brinkhoff, H-P Kriegel and B. Seeger, Proc. SIGMOD, 1993.
  • Spatial Joins using Seeded Trees ”, M-L Lo and C. V. Ravishankar, Proc. SIGMOD,
  • Spatial Hash-Joins ”, M-L Lo and C. V. Ravishankar, Proc. SIGMOD, 1996.
  • Size Separation Spatial Joins ”, N. Koudas and K. C. Sevcik, Proc. SIGMOD, 1997.

3

Spatial Join Definition

  • Unlike equi-join in relational DBMS, intersection join!
  • Join result: (x2, y1), (x2, y2)

x

x

y

y

y

Naïve approach

  • For every object in X, check against every object in Y.
  • O(n^2 ) -- not efficient.
  • Idea to improve: cluster together objects that are spatially close to each other.

7

Seeded Tree

  • An R-tree like structure, built on a set of spatial objects based on an existing R-tree.
  • The seeded-tree copies the top levels of the existing R-tree, thus forcing the index to have the same clustering as the existing tree.
  • Benefit: improves join efficiency.
    1. Spatial Join using Seeded Trees

Building A Seeded Tree

  • Seeding phase : copy the top k levels of the existing R-tree. Bottom level contains a set of slots.
  • Growing phase : insert objects into slots. Each slot may grow to a sub-tree.
  • Clean-up phase : empty slots are erased; slot MBRs are adjusted. 2. Spatial Join using Seeded Trees

9

Seeded Tree base Spatial Join

  • Used when there exists an R-tree on X, but not on Y.
  • Build a seeded tree on Y, then join with the R-tree on X.
  • Seed level filtering : in the growing phase, if an object does not intersect any seed-level slot, no need to insert! 2. Spatial Join using Seeded Trees

Spatial Hash Join

  • Motivation: if all objects fit in memory, then trivial. But size of input is large.
  • Basic idea: partition input into smaller buckets that fit in memory. 3. Spatial Hash Join
  • Difficulty: an object may intersect multiple partitions!

13

Spatial Hash Join

  • Assign every object in X to a single bucket. Bucket extent may increase accordingly.
  • Using the same partitioning, assign objects in Y to all buckets that it intersects.
  • Join corresponding bucket pairs (a bucket of X joins with a single bucket of Y). 3. Spatial Hash Join

™ Better than both previous approaches!

Size Separation Spatial Join

Overview

  • Partition both object sets by object size into L+1 levels.
  • Each level is logically partitioned.
  • Join the partitions synchronously.
    1. Size Separation Spatial Join

15

Partitioning

  • Level l contains 2 l^ regular grid cells.
  • If an object intersects the grid of level l , assign it to level l -1.
  • Result: partition (roughly) by size; each object belongs to a single partition at some level. 4. Size Separation Spatial Join

Level 0

Level 1

Level 2

Joining

  1. Size Separation Spatial Join

Level 0

Level 1

Level 2

  • A bucket in Y needs to join with a single bucket of X at each level.