Data Mining and Multi-agent Integration (eBook)

Longbing Cao (Herausgeber)

eBook Download: PDF

2009 | 2009
XIV, 334 Seiten
Springer US (Verlag)
978-1-4419-0522-2 (ISBN)

Infos

Data Mining and Multi agent Integration aims to re?ect state of the art research and development of agent mining interaction and integration (for short, agent min ing). The book was motivated by increasing interest and work in the agents data min ing, and vice versa. The interaction and integration comes about from the intrinsic challenges faced by agent technology and data mining respectively; for instance, multi agent systems face the problem of enhancing agent learning capability, and avoiding the uncertainty of self organization and intelligence emergence. Data min ing, if integrated into agent systems, can greatly enhance the learning skills of agents, and assist agents with predication of future states, thus initiating follow up action or intervention. The data mining community is now struggling with mining distributed, interactive and heterogeneous data sources. Agents can be used to man age such data sources for data access, monitoring, integration, and pattern merging from the infrastructure, gateway, message passing and pattern delivery perspectives. These two examples illustrate the potential of agent mining in handling challenges in respective communities. There is an excellent opportunity to create innovative, dual agent mining interac tion and integration technology, tools and systems which will deliver results in one new technology.

Inhalt

Data Mining and Multiagent Integration 2
Preface 5
Foreword 8
Contents 10
Introduction to Agents and Data Mining Interaction 13
Introduction to Agent Mining Interaction and Integration 14
1.1 Introduction 14
1.2 Driving forces of agent mining interaction and integration 16
1.2.1 Challenges in agent disciplines 16
1.2.2 Challenges in data mining disciplines 18
1.2.3 Mutual challenges in agent and mining 20
1.3 Complementary essence and interaction potential of agentsand data mining 21
1.4 A Disciplinary Framework of Agent and Mining Interactionand Integration 23
1.4.1 Evolution process and characteristics 23
1.4.2 Agent-mining interaction framework 25
1.4.3 Theoretical underpinnings for agent mining 28
1.4.4 Agent mining lifecycle and process 29
1.5 Agent-Driven Distributed Data Mining 30
1.5.1 The challenges of distributed data mining 30
1.5.2 The needs of agent-driven distributed data mining 31
1.5.3 Research issues in agent driven data mining 32
1.6 Data Mining-Driven Agents 33
1.6.1 The challenges of data mining-driven agents 33
1.6.2 The needs of data mining-driven agents 35
1.6.3 Research issues in data mining driven agents 36
1.7 Mutual Issues in Agent Mining 37
1.7.1 The need to study common issues for agent mining 37
1.7.2 Mutual research issues in agent mining 38
1.8 Applications and case studies 39
1.8.1 Applications 39
1.8.2 Case Studies: Developing Actionable Trading Agents 40
Evolutionary Trading Agents for Parameter Optimization 40
1.8.3 F-Trade: An agent-mining symbiont 41
1.8.4 Agent Academy 42
1.9 Trends and directions 43
1.10 Agent mining community development 44
1.10.1 Fast progression in fostering the community 44
1.10.2 Research resources on agent-mining interaction 44
1.11 Conclusions 45
Acknowledgements 45
References 45
Towards the Integration of Multiagent 48
2.1 Overview of Agents/Multiagent Systems and Data Mining Integration 48
2.2 RelatedWork 49
2.3 BioAgents 51
2.4 MADIK 52
2.5 Evaluation and FutureWork 53
References 55
Agent-Based Distributed Data Mining: A Survey 58
3.1 Introduction 58
3.2 Why Agents 58
3.3 Agent-Based Distributed Data Mining 60
3.4 Interaction and Integration 62
3.5 Open Issues and Trends 65
References 67
Data Mining Driven Agents 70
Exploiting Swarm Behaviour of Simple Agents for ClusteringWeb Users’ Session Data 71
4.1 Introduction 71
4.2 Web Usage Mining 73
4.2.1 Web Session Clustering 73
4.2.2 Session Identification 73
4.2.3 Web Session Clustering Techniques 74
4.3 Swarm Intelligence 76
4.3.1 Particle Swarm Optimization 76
4.3.2 PSO Based Web Usage Clustering 78
4.4 Experimental Results 80
4.4.1 Data, Pre-processing and Usage Statistics 80
4.4.2 Clustering Results 81
4.5 RelatedWork 83
4.6 Conclusion and FutureWork 84
References 84
Mining Temporal Patterns to Improve Agents Behavior: Two Case Studies 86
5.1 Introduction 86
5.2 Mining Temporal Patterns from Sequences of Events 87
5.3 Agents that Learn from Other Agents 88
5.3.1 The Observing Phase 90
5.3.2 The Learning Phase 91
5.3.3 The Application Phase 91
5.3.4 An Experiment 93
5.4 Agents that Learn from Their Own Behavior 94
5.4.1 The CTS Cognitive Agent 94
5.4.2 The Observation Phase 95
5.4.3 The Learning Phase 96
5.4.4 The Application Phase 96
5.4.5 Testing the New CTS in RomanTutor 97
5.5 Conclusion 100
Acknowledgements 100
References 100
A Multi-Agent System for Extracting and Analysing Users’ Interaction in a Collaborative Knowledge Management System 102
6.1 Introduction 102
6.2 RelatedWork 103
6.3 The proposal Context: the KnowCat System and its ClientMonitor 104
6.4 A Multi-Agent System for Extracting and Analysing Users’Interaction 105
6.4.1 Data Extractor Agent 106
6.4.2 Organised Behaviour Interpreter Agent 107
6.5 Experimentation and Results 108
6.6 Conclusions and FutureWork 110
Acknowledgements 111
References 111
Towards Information Enrichment through Recommendation Sharing 112
7.1 Introduction 112
7.2 Prior and RelatedWork 114
7.3 Ecommerce-Oriented Distributed Recommender System 116
7.3.1 Interaction Protocol 121
7.4 Peer Profiling and Selection 123
7.4.1 System Formalization for EDRS 123
7.4.2 User Clustering 124
7.4.3 Recommender Peer Profiling 124
7.4.4 Recommender Peer Selection 126
7.4.4.1 Gittins Indices 126
7.4.4.2 Selection Strategy for EDRS 128
7.5 Experiments and Evaluation 128
7.5.1 Data Acquisition 128
7.5.2 Experiment Setup 129
7.5.2.1 Constructing Recommender Peers 129
7.5.2.2 Evaluation Metrics 130
7.5.2.3 Benchmarks for the Peer Profiling and Selection Strategy 131
7.5.3 Experimental Results 131
7.6 Conclusions 134
References 134
A Multiagent-based Intrusion Detection System with the Support of Multi-Class Supervised 136
8.1 Introduction 136
8.2 ExistingWork 138
8.3 The Proposed DMAS-IDS Architecture 139
8.3.1 Agent Architecture 140
8.3.1.1 Host Layer 140
8.3.1.2 Classification Layer 141
8.3.1.3 Manager Layer 141
8.3.1.4 Agent Communication 141
8.3.2 Principal Component Classifier (PCC) 142
8.3.3 Collateral Representative Subspace Projection Modeling(C-RSPM) 144
8.3.4 Policy Derivation at Classification Agents 145
8.4 Experimental Setup 146
8.5 Results and Analysis 148
8.6 Conclusion 149
References 150
AutomaticWeb Data Extraction Based on Genetic Algorithms and Regular Expressions 152
9.1 Introduction 152
9.2 Genetic Algorithms and Its Application in Wrappers andRegular Expressions 153
9.2.1 Genetic Algorithms 153
9.2.2 Regular Expressions 155
9.3 How Agents Support Data Mining: Variable LengthPopulation 155
9.3.1 Macroevolution 156
9.3.2 Microevolution 157
9.4 Composition of Basic Regex 158
9.5 Experimental Evaluation 159
9.5.1 Results: Regex Evolution 160
9.5.2 Results: Regex Composition 160
9.6 Conclusions 163
Acknowledgements 163
References 163
Establishment and Maintenance of a Knowledge Network by Means of Agents and Implicit Data 164
10.1 Introduction 164
10.2 The SKC Analysis Module 167
10.3 The SKC Network Module 168
10.4 Experiments Carried Out 172
10.5 Conclusions 174
Acknowledgements 175
References 175
Equipping Intelligent Agents with Commonsense Knowledge acquired from Search 176
11.1 Motivation 176
11.2 Goal Mining 178
11.3 Explicit User Goals: Definition and Agreeability 179
11.3.1 Results of Human Subject Study 179
11.4 Case Study 180
11.5 Results 181
11.6 Conclusions 183
Acknowledgements 183
References 184
A Multi-Agent Learning Paradigm for Medical Data Mining DiagnosticWorkbench 186
12.1 Introduction 186
12.2 Multi-Agent Learning Paradigm of i+DiaMAS 188
12.3 Data Preprocessing Agent (AgPre) 189
12.4 Data Mining Agent (AgDM) 191
12.5 Evaluation 193
12.6 Conclusion 194
References 194
Agent Driven Data Mining 196
The EMADS Extendible Multi-Agent Data Mining Framework 197
13.1 Introduction 197
13.2 RelatedWork 199
13.3 The EMADS Conceptual Framework 199
13.3.1 EMADS End User Categories 200
13.4 The EMADS Implementation 201
13.4.1 EMADS Wrappers 202
13.5 EMADS Operations 203
13.5.1 Meta ARM (Association Rule Mining) Scenario 203
13.5.1.1 Dynamic Behaviour of EMADS for Meta ARM Operations 203
13.5.1.2 Experimentation and Analysis 204
13.5.2 Classifier Generation Scenario 205
13.5.2.1 Experimentation and Analysis 206
13.6 Conclusions 207
References 208
A Multiagent Approach to Adaptive Continuous Analysis of Streaming Data in Complex Uncertain Environments 209
14.1 Introduction 209
14.1.1 Problem Definition 209
14.1.2 Related Work 211
14.2 Continuous Online Unsupervised Learning in ComplexUncertain Environments 212
14.2.1 Market-based Algorithm of Continuous AgglomerativeHierarchical Clustering 212
14.2.2 Agent Decision-making Model 214
14.3 Experimental Analysis 216
14.3.1 Datasets 216
14.3.2 Experimental Results 216
14.4 Summary, Conclusion and FutureWork 221
14.4.1 Summary 221
14.4.2 Conclusions and Future Directions 222
Appendix 224
References 226
Multiagent Systems for Large Data Clustering 227
15.1 Introduction 227
15.1.1 Motivation and Why ADMI 228
15.1.2 Current Literature and Proposed Approach 229
15.1.2.1 Agents Supporting Data Mining 229
15.1.2.2 Data Mining Supporting Agents 229
15.1.2.3 Proposed Approach 230
15.2 Scheme-1: Multiagent Based Clustering using Divide andConquer Approach 231
15.2.1 Motivation 231
15.2.2 Description of Handwritten Digit Data and Preliminary Analysis 232
15.2.3 Description of Intrusion Detection Data and Preliminary Analysis 232
15.2.4 Proposed Multiagent System for the Divide and ConquerMethod 235
15.2.4.1 Experimental Results for Handwritten Digit Data 235
15.2.4.2 Experimental Results for Network Intrusion Detection Data 236
15.2.5 Summary 237
15.3 Scheme-2: Multiagent Based Clustering Using DataDependent Schemes 238
15.3.1 Motivation 238
15.3.2 Choice of Prototype Selection Algorithm 239
15.3.3 Proposed Multiagent System for Large Data Clusteringbased Data Dependent Scheme 240
15.3.3.1 Experimental Results with Handwritten Digit Data 241
15.3.3.2 Experimental Results with Network Intrusion Detection Data 242
15.3.4 Summary 243
15.4 Summary and Further work 243
References 244
A Multiagent, Multiobjective Clustering Algorithm 247
16.1 Introduction 247
16.2 RelatedWork 248
16.3 MACC – Multi Ant Colony Clustering Algorithm 250
16.4 Experiments and Results 254
16.5 Conclusion and FutureWork 256
Acknowledgements 256
References 256
Integration of Agents and Data Mining in InteractiveWeb Environment for Psychometric Diagnostics 258
17.1 Introduction 258
17.2 Interactive Environment for Psychometrics Diagnostics 260
17.3 Collecting Information about Users 261
17.4 Guiding Users Trough Diagnostic Tests and TrainingPrograms 262
17.5 Interpretation and Visualization of Results 264
17.6 Cognitive Profile of User 265
17.7 Analysis of Psychometric Data 265
17.8 Helping Psychologist in Their Activities 268
17.9 Agent’s Role in Environment for Psychometric Diagnostic 268
17.10 Conclusion 270
References 270
A Multi-Agent Framework for Anomalies Detection on Distributed Firewalls Using Data Mining Techniques 273
18.1 Introduction 273
18.2 Background 274
18.3 Data Mining Techniques and Their Application on Firewalls 274
18.4 Integration of Agents and Data Mining 276
18.5 The Principal Contribution 278
18.5.1 Intra-Firewalls Anomalies Detection Model 278
18.5.2 Inter-Firewalls Anomalies Detection Model 279
18.6 Case Study 279
18.6.1 Intra-Firewalls Anomalies Detection Results 280
18.6.2 Inter-Firewalls Anomalies Detection Results 280
18.7 Conclusion 281
References 283
Competitive-Cooperative Automated Reasoning from Distributed and Multiple Source of Data 285
19.1 Introduction 285
19.2 RelatedWork 286
19.3 The Proposed Approach 286
19.3.1 ARM Agent Behavior 288
19.3.1.1 Apriori Algorithm 289
19.3.1.2 Game theory approach 289
19.3.2 Inference Agent Behavior 291
19.3.3 Response Agent Behavior 292
19.4 System Run Sample 293
19.5 Conclusion and FutureWork 295
Acknowledgements 295
References 295
Normative Multi-Agent Enriched Data Mining to Support E-Citizens 297
20.1 Introduction 297
20.2 Theoretical issues – A Multi-Agent System vs. Norms 299
20.3 A Normative Multi-Agent Enriched Data miningArchitecture and Ontology Frameworks 301
20.4 Case Study – A Multi-Agent System Architecture 304
20.5 Concluding Remarks 308
References 309
Chapter 21 311
21.1 Introduction 311
21.2 Background and RelatedWork 312
21.2.1 Interactive Museums and Virtual Learning Communities 312
21.2.2 Data Mining 313
21.2.3 Related Works on Data Mining and Virtual Community 314
21.3 CV-Muzar (Augusto Ruschi Zoobotanical Museum VirtualCommunity of the University of Passo Fundo) 314
21.3.1 Sub-Communities Formations Assisted by a MultiagentSystem 315
21.3.1.1 Investigating Society of Sub-Community (SIS-C) 315
21.3.2 Initial Evaluation 318
21.3.1.2 Investigating Society of Participants (SIP) 316
21.4 Final Considerations and FutureWork 319
References 320
Agent based Video Contents Identification and Data Mining UsingWatermark based Filtering 321
22.1 Introduction 321
22.2 Multagent Integration and Data Mining Concept 322
22.3 Watermark Embedding for Contents Identification 323
22.4 Agent based Content Blocking and Tracing 325
22.4.1 Contents Blocking Agent 325
22.4.2 Monitoring Agent 325
22.4.3 Data Mining and Reporting Process 327
22.5 Experimental Evaluations 328
22.6 Conclusion 329
References 330
Index 331

Produkt-Details

• Erscheint lt. Verlag: 25.7.2009
• Zusatzinfo: XIV, 334 p.
• Verlagsort: New York
• Sprache: englisch
• Themenwelt: Informatik ► Datenbanken ► Data Warehouse / Data Mining
• Themenwelt: Informatik ► Theorie / Studium ► Künstliche Intelligenz / Robotik
• Schlagworte: AAMAS • agent-enriched data mining • algorithm • algorithms • automated reasoning • classification • Clustering • currentjm • Data Mining • filtering • Genetic algorithms • knowledge management • learning • Multi-agent Systems • Regular Expressions
• ISBN-10: 1-4419-0522-7 / 1441905227
• ISBN-13: 978-1-4419-0522-2 / 9781441905222