Publications

Detailed Information

Graph Indexing for Label-Constrained Reachability Queries : 제한된 라벨 도달 질의를 위한 그래프 인덱싱 기법

DC Field Value Language
dc.contributor.advisorSrinivasa Rao Satti-
dc.contributor.authorMohammadsadegh Najafi-
dc.date.accessioned2020-10-13T02:59:02Z-
dc.date.available2020-10-13T02:59:02Z-
dc.date.issued2020-
dc.identifier.other000000161949-
dc.identifier.urihttps://hdl.handle.net/10371/169365-
dc.identifier.urihttp://dcollection.snu.ac.kr/common/orgView/000000161949ko_KR
dc.description학위논문 (석사) -- 서울대학교 대학원 : 공과대학 컴퓨터공학부, 2020. 8. Srinivasa Rao Satti.-
dc.description.abstract계속해서 발전하는 세계에서 우리는 간선에 레이블이 포함된 그래프로 표현된 소셜 네트워크, 세만틱 웹 및 인용 네트워크 데이터를 계속해서 접하게 된다.
이렇게 그래프 형태로 구축된 데이터 내의 지식은 기저의 그래프를 질의함으로서 영향력을 발휘한다.
기본적으로 수행되는 그래프 질의는 주어진 소스 정점과 타켓 정점간의 도달 가능성을 판별한다.
또한 해당 질의는 간선에 라벨이 부여된 그래프 형태에 대해서 대상 라벨 집합을 제한하는 형태로도 주로 수행된다.
Valstar 등[SIGMOD-2017]에 의해 상기 문제가 제시되었으며, 그들은 해당 라벨 제한 질의를 빠르게 지원하는 landmark indexing이라는 방식을 제안하였다.
본 논문을 통해 우리는 간편하고 실제적이며 공간 효율적으로 처리할 수 있는 제한된 라벨 도달 질의 수행 방식을 소개한다.
다양한 종류의 그래프에 대한 실험을 통하여 우리가 제시한 질의 알고리즘이 상당한 수준의 공간 효율성을 보임을 확인한다.
또한 landmark indexing과 비교하였을 때 제안된 처리 방식의 질의 처리 시간 및 사용되는 공간이 개선되었음을 보인다.
-
dc.description.abstractIn our ever-growing world, we are faced with graph data in social networks, semantic webs, or a citation network stored in an edge-labeled graph. A good deal of knowledge in such a graph-structured data can be leveraged by querying the underlying graph. A fundamental query determines reachability between a given source and a target. The reachability query in the edge-labeled graph often also involves constraining the label set. This problem has recently been addressed by Valstar et al.~[SIGMOD-2017], who proposed an approach called Landmark Indexing to support label-constrained queries faster. In this work, we introduce simple, practical, and space-efficient solutions for efficiently evaluating label-constrained reachability (LCR) queries. Experimental results on both real and synthetic graphs demonstrate significant space efficiency benefits of our solutions. In fact, our schemes improve upon both the query time and space usage in comparison with landmark indexing.-
dc.description.tableofcontentsChapter 1 Introduction 1
1.1 Rechability Concept 1
1.2 Label Constrained Reachability 2
1.3 The Problem 4
1.4 State of the Art 5
1.5 Main Contributions 5
Chapter 2 Literature Analysis 7
2.1 Landmark Indexing [35] 7
2.1.1 Constructing Indices for Landmarks 8
2.1.2 Query Procedure 9
2.1.3 Accelerating False Queries 10
2.1.4 Experiments on the Landmark Indexing Procedure 11
2.2 Transitive-closure Method [46] 13
2.2.1 Finding Single-source Transitive Closure 14
2.2.2 Graph Transformation to Direct Acyclic Graph 15
2.2.3 Augmented DAG Procedure 17
2.2.4 Results 18
2.3 Shortest Path Label-Constrained [6] 19
2.3.1 Results of Shortest Path Label-constrained 21
2.3.2 LCR Application 22
2.4 Tree-based Index Framework [18] 23
2.5 Approximate Regular-simple-path Reachability [36] 24
2.6 Introduction to General Reachability Queries [42] 25
2.7 BFS Optimization [5] 26
2.8 Directed Acyclic Graph Reachability 26
Chapter 3 Methods Derived from All Label Combinations 28
3.1 All Labels Combinations (ALC) 29
3.1.1 Decomposing Algorithm 29
3.1.2 Query Processing Algorithm 31
3.1.3 Time and Space Analysis 31
3.2 Incorporating Strongly Connected Component (SCC) Decomposition in ALC Algorithm (ALC+SCC) 33
3.2.1 Decomposition Algorithm 33
3.2.2 Query Algorithm 34
3.2.3 Time and Space Analysis 37
3.3 ALC+SSC Combined With Landmark Indexing (ASL) 37
3.3.1 Landmark Indexing Concept 37
3.3.2 ASL Procedure 39
3.3.3 Query Algorithm 40
3.3.4 Time and Space Analysis 41
3.4 ALC Space Reduction 41
3.5 Supporting Updates 43
3.5.1 Adding an Edge 44
3.5.2 Removing an Edge 44
3.5.3 Updating the Label of an Edge 45
3.5.4 Adding a Node 45
3.5.5 Removing a Node 45
3.6 Accelerating Construction 45
3.7 Extension 46
3.7.1 Distributed Graphs 46
3.7.2 Distributed Reachability 47
3.7.3 Distributed Reachability for LCR Queries 47
3.7.4 Query Processing 48
Chapter 4 Methods Derived from Unlabeled Graphs 49
4.1 Unlabeled Graphs 49
4.1.1 NLG Procedure 50
4.1.2 NLG Decomposing 51
4.1.3 NLG Query Processing 51
4.1.4 Space Usage and Query Answering Time Complexity 53
4.2 Disjoint Reachability 53
4.2.1 Disjoint Reachability Procedure 53
4.2.2 Disjoint Reachability Construction 55
4.2.3 Disjoint Reachability Query - Index Traversal 56
4.2.4 Disjoint Reachability Space Usage and Query Answering Time Complexity 57
4.3 Distributed Reachability Application 60
Chapter 5 Experiments 61
5.1 ALC, ALC+SCC, and ASL Versus Landmark Indexing 61
5.2 Data Sets 62
5.3 Queries 64
5.4 Performance Analysis 65
5.4.1 Query Time 66
5.4.2 Memory Usage 69
5.4.3 Construction Time 71
5.5 ASL Speed-up 73
5.6 Space Reduction Performance 73
5.7 Acceleration Performance 78
5.8 Supporting Updates 79
5.9 Performance of NLG and DR Versus Other Algorithms 80
5.10 Scalability 83
5.11 Applying Compression Techniques 84
Chapter 6 Conclusions 86
Bibliography 88
-
dc.language.isoeng-
dc.publisher서울대학교 대학원-
dc.subjectLabel Constrained Reachability Queries-
dc.subjectEdge Labeled Graph-
dc.subjectDisjoint Reachability-
dc.subjectBFS-
dc.subject제한된 라벨 도달 질의-
dc.subject간선 레이블 그래프-
dc.subject개별적 도달 가능성-
dc.subject너비 우선 탐색-
dc.subject.ddc621.39-
dc.titleGraph Indexing for Label-Constrained Reachability Queries-
dc.title.alternative제한된 라벨 도달 질의를 위한 그래프 인덱싱 기법-
dc.typeThesis-
dc.typeDissertation-
dc.contributor.AlternativeAuthor무함마드사데흐-
dc.contributor.department공과대학 컴퓨터공학부-
dc.description.degreeMaster-
dc.date.awarded2020-08-
dc.identifier.uciI804:11032-000000161949-
dc.identifier.holdings000000000043▲000000000048▲000000161949▲-
Appears in Collections:
Files in This Item:

Altmetrics

Item View & Download Count

  • mendeley

Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.

Share