1. Introduction to Spatial Database 2003-07-26 김동오 건국대학교 데이타베이스연구실

2. Content  Overview  Who can benefit from spatial data menagement?  GIS and SDBMS  Three classes of users for spatial database  A Stroll Through Spatial Databases  Summary

3. Overview  공간 데이타나 공간 질의를 다루기 위한 데 이타베이스의 필요

4. Who can benefit from spatial data menagement?  Mobile phone user  Where is the nearest gas station?  Is there a pet-food vendor on my way home?  Army field commander  Has there been any significant enemy troop movement since last night?  Emergency service  Where is the person calling for help located?  Where is the best route to reach her? ...

5. GIS and SDBMS  SDBMS  SDBMS are designed to handle very large amounts of spatial data stored on secondary devices, using specialized indices and query- processing techniques  SDBMS provides simpler operations on a set of objects and sets of layers  GIS  Geographic Information System(GIS) is the principal technology motivating interest in Spatial Database Management Systems(SDBMSs)  GIS provides a convenient mechanism for the analysis and visualization of geographic data  GIS provides a rich set of operations over few objects and layers  GIS can be built as the front-end of an SDBMS  An efficient SDBMS can greatly increase the efficiency and productivity of a GIS

6. Three classes of users for spatial database  Specialized data management  Analysis of spatial data  GIS middleware - ESRI's Spatial Data Engine  Spatial operation & Representing geographic information  LBS - use GPS and mobile devices 1980s 1990s 2000s Time GISystems GIScience GIServices

7. A Stroll Through Spatial Databases  Space taxonomy and data models  Query Language  Query processing  File organization and indices  Query optimization  Data mining

8. Space taxonomy and data models  Space taxonomy  공간을 구성 가능한 다중 서술  ex) 연산을 위한 다양한 공간 타입  ( Topological - Adjacent ), (Network - Shortest-path) ( Directional - North-of ), (Euclidean - Distance)  Data model  공간에 관련된 객체를 표현하고 식별하기위한 규칙이나 규칙의 집합  ex) 서울시는 2 차원 평면, 1 차원 선, 무수히 많은 점들의 집합 또는 점으 로 표현  Object model  정형적인 공간 개체를 표현하는 이상적인 것  Conceptual model  Map onto the computer using the Vector data structure  Vector data structure  Map region into polygon, line, point

9. Query processing  Query  Single scan query  Ex) Window query  Multi scan query  Ex) join  Range query  use filter-refine paradigm FILTER REFINE

10. Query processing  An algorithm to process the filter step of a spatial-join query  Plane sweep technique  Move a sweep line from left to right and stop at the first entry in R ∪ S. This rectangle is T.  Search through the sorted rectangles of S until arriving at the first rectangle S f such that S f.xl > T.xu. Search S j in sorted rectangles of S for S j < S f.  If (T.yl, T.yu) ∩ (S j.yl, S j.yu) is not null, then the S j intersects T. are part of the join result. Remove T from the set R ∪ S  Move the sweep line across the set R ∪ S until it reaches the next rectangle entry. Now proceed as in step 2 and 3.  Stop when the R ∪ S = Ø R4 S2 R1 S1 S3 R2 R3 sweep line ( T.xl, T.yl ) ( T.xu, T.yu ) R4S2S1R1S3R2R3

11. File organization and indices  Minimizing the sum of the compution and the I/O time  Search in the index  Sort and Order  Row-ordered or Z-ordered  B-tree  R-tree CPU Main Memory CPU DISK 1234 5678 9101112 13141516 1256 3478 9101314 11121516

12. Query optimization  Join query  Select before join or join before select ?  "Find all senators who serve a district of area greater than sqare miles and who own a business within the district"  A range query and a spatial join query  "Name all senators who serve a district of area greater than 300 square miles"  "Find all senators who own a business located in the district they serve"

13. 연산 횟수 Query optimization  "Name all senators who serve a district of area greater than 300 square miles"  "Find all senators who own a business located in the district they serve"  spatial join query -> range query  range query -> spatial join query namedistrict 김동오한국 이동진남해 손훈수대전 강홍구의정부 namebusinesslocation 김동오식당여수 손훈수식당서울 장유정미용실서울 namedistrict 김동오한국 손훈수대전 namebusinesslocation 김동오식당여수 손훈수식당대전 장유정미용실서울 namedistrict 김동오한국 이동진남해 손훈수대전 강홍구의정부 namedistrictbusinesslocation 김동오한국식당여수 4 + ( 2 * 3 ) *α = 10 ( 4 * 3 ) *α + 1 = 13

14. Summary  공간 데이타 관리는 다양한 분야에서 사용 되며 과학적으로 중요한 역할을 담당한다.  도시 계획, 자연 자원 관리, 지구 기후 변화 측 정 등  사용자 별 공간 데이타 관리의 이점  Business user : 지리 정보를 활용한 마케팅, 점 포 분할 등 중요한 이차 정보로 활용 가능  Scientific user : 환경 및 자원 연구의 전문화 가 능  Personal user: 지역적 개인화가 가능