MINING GRAPH DATA EDITED BY. Diane J. Cook School of Electrical Engineering and Computei' Science Washington State University Puliman, Washington

Transcription

1 MINING GRAPH DATA EDITED BY Diane J. Cook School of Electrical Engineering and Computei' Science Washington State University Puliman, Washington Lawrence B. Holder School of Electrical Engineering and Computer Science Washington State University Puliman, Washington BICINTlUNIfl Jl O 7 l WILEY: ; <n\ ftlcintinhial WILEY-INTERSC1ENCE A JOHN WILEY & SONS, INC., PUBLICATION

2 CONTENTS Preface Acknowledgments Contributors xiii xv xvn 1 INTRODUCTION 1 Lawrence B. Holder and Diane J. Cook 1.1 Terminology Graph Databases 3 i.3 Book Overview 10 References 11 Part I GRAPHS 15 2 GRAPH MATCHING EXACT AND ERROR-TOLERANT METHODS AND THE AUTOMATIC LEARNING OF EDIT COSTS 17 Horst Bunke and Michel Neuhaus 2.1 Introductkm Definitions and Graph Matching Methods Learning Edit Costs Experimental Evaluation Discussion and Conelusions 31 References 32 3 GRAPH VISUALIZATION AND DATA MINING 35 Walter Didimo and Giuseppe Liotta 3.1 Introduction Graph Drawing Techniques Examples of Visualization Systems 48 vii

3 viii CONTENTS 3.4 Conclusions 55 Rer'erences 57 4 GRAPH PATTERNS AND THE R-MAT GENERATOR 65 Deepayan Chakrabarti and Christas Faioutsos 4.1 Introdüction Background and Related Work NetMine and R-MAT Experiments Conciusions 86 References 92 Part II MINING TECHNIQUES 97 5 DISCOVERY OF FREQUENT SUBSTRUCTURES 99 Xifeng Yan and Jiawei Han 5.1 Introdüction Preliminary Concepts Apriori4)ased Approach Pattern Growth Approach Variant Substrueture Patterns Experiments and Performance Study Conclusions 112 References FINDING TOPOLOGICAL FREQUENT PATTERNS FROM GRAPH DATASETS 117 Michihiro Kuramochi and George Karypis 6.1 Introdüction Background Definitions and Notation Frequent Pattern Discovery from Graph Datasets Problem Definitions FSG for the Graph-Transaction Setting S-GKAM tbr the Single-Graph Setting GREW Scalable Frequem Subgraph Discovery Algorithm Related Research Conclusions 151 References UNSUPERVISED AND SUPERVISED PATTERN LEARNING IN GRAPH DATA 159 Diane J. Cook, Lawrence B. Holder, and Nikhil Keikar 7.1 Introdüction 159

4 CONTENTS ix 7.2 Mining Graph Data Using Subdue Comparison to Other Graph-Based Mining Algorithms Comparison to Frequent Substructure Mining Approaches Comparison to ILP Approaches Conclusions 179 References GRAPH GRAMMAR LEARNING 183 Istvan Jonyer 8.1 Introduction Related Work Graph Grammar Learning Empirical Evaluation Conclusion 199 References CONSTRUCTING DECISION TREE BASED ON CHUNKINGLESS GRAPH-BASED INDUCTION 203 Kouzou Ohara, Phu Chien Nguyen, Akira Mogi, Hiroshi Motoda, and Takashi Washio 9.1 Introduction Graph-Based Induction Revisited Problem Caused by Chunking in B-GBI Chunkingless Graph-Based Induction (Cl-GBI) Decision Tree Chunkingless Graph-Based Induction (DT-C1GBI) Conclusions 224 References SOWIE LINKS BETWEEN FORMAL CONCEPT ANALYSIS AND GRAPH MINING 227 Michel Liquiere 10.1 Presentation Basic Concepts and Notation Formal Concept Analysis Extension Lattice and Description Lattice Give Concept Lattice Graph Description and Galois Lattice Graph Mining and Formal Propositionalization Conclusion 249 References 250

5 X CONTENTS 11 KERNEL METHODS FOR GRAPHS 253 Thomas Gärtner, Tamäs Horvath, Quoc V, Le, Alex J. Smola, and Stefan Wrobel 11.1 Introduction 11.2 Graph Classification 11.3 Vertex Classification 11.4 Conclusions and Future Work References KERNELS AS LINK ANALYSIS MEASURES Masashi Shimbo and Takahiko Ito 12.1 Introduction 12.2 Preliminaries 12.3 Kernel-based Unified Framework for Importance and Relatedness 12.4 Laplacian Kernels as a Relatedness Measure 12.5 Practical Issues 12.6 Related Work 12.7 Evaluation with Bibliographie Citation Data 12,8 Summary References ENTITY RESOLUTION IN GRAPHS 311 Indrajit Bhattacharya and Lise Getoor 13.1 Introduction Related Work Motivating Example for Graph-Based Entity Resolution Graph-Based Entity Resolution: Problem Formulation Similarity Measures for Entity Resolution Graph-Based Clustering for Entity Resolution Experimental Evaluation 333!3.8 Conclusion 341 References 342 Part III APPLICATIONS MINING FROM CHEMICAL GRAPHS 347 Takashi Okada 14.1 Introduction and Representation of Molecules Issues for Mining CASE: A Prototype Mining System in Chemistry Quantitative Estimation Using Graph Mining Extension of Linear Fragments to Graphs 362

6 CONTENTS xi 14.6 Combination of Conditions Concluding Remarks 375 References UNIFIED APPROACH TO ROOTED TREE MINING: ALGORITHMS AND APPLICATIONS 381 Mohammed Zaki 15.1 Introducüon Preliminaries Related Work Generaüng Candidate Subtrees Frequency Computation Counting Distinct Occurrences The SLEUTH Algorithm Experimental Results Tree Mining Applications in Bioinformatics Conclusions 409 References DENSE SUBGRAPH EXTRACTION 411 Andrew Tomkins and Ravi Kumar Introduction Related Work Finding the densest subgraph Trawling Graph Shingling Connection Subgraphs Conclusions References SOCIAL NETWORK ANALYSIS 443 Sherry E. Marcus, Melanie Moy, and Thayne Cojfman 17.1 Introduction Social Network Analysis Group Detection Terrorist Modus Operandi Detection System Computational Experiments Conclusion 467 References 468 Index 469