Embed Size (px)
Citation preview
지능정보시스템연구실
황 지 민
I N D E X
10.1 Introduction
10.2 A Taxonomy of Distributed IR System
10.3 Data Partitioning
10.4 Parallel IR
10.5 Cluster-based IR
10.1 Introduction
10.1 4
10.2 A Taxonomy of
Distributed IR Systems
10.2 6
표준 검색 병렬검색 SIMD 병렬검색 MIMD
n/a 클러스터 기반
검색 지역 통합 검색
n/a 분산 검색(P2P) 통합(Federated) 검색
(P2P)
10.2 7
10.2 8
10.2 9
10.3 Data Partitioning
10.3 11
색인 항목
문 헌
10.3 12
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 13
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 14
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 15
10.3 16
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 17
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
𝑤𝑐,𝑖𝑗 = 𝑓𝑐,𝑖𝑗 × 𝐼𝐷𝐹𝑐,𝑖 컬렉션내 용어 𝑘𝑖 의 가중치
컬렉션의 모든 문헌들 내에 있는 용어 𝑘𝑖 의 전체 출현빈도
역 컬렉션 빈도
𝐼𝐷𝐹𝑐,𝑖 = 𝑙𝑜𝑔𝑁𝑐
𝑛𝑐,𝑖 𝑛𝑐,𝑖: 용어 𝑘𝑖 가 출현하는 컬렉션의 수
𝑁𝑐: 전체 컬렉션의 수
10.3 18
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 19
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 20
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 21
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 22
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 23
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 24
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 25
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.3 26
1. Collection
partitioning
2. Collection
selection
3. inverted index
partitioning
4. Partitioning
other indexes
10.4 Parallel IR
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 28
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 29
∅ = 1 일때, 이상적인 효율
𝑆:속도향상, 𝑁:프로세서의수
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 30
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 31
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 32
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 33
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 34
1. introduction
2. MIMD
architectures
3. SIMD
architectures
10.4 35
10.5 Cluster-based IR
10.5 37
10.5 38
𝐿𝐵 = max𝑖=1
𝑛|𝑙𝑜𝑎𝑑𝑖 − 𝑙|
𝑙
10.5 39
END
