Bisociative Knowledge Discovery

Description

Cover Front matter Title Foreword Table of Contents Part I: Bisociation Towards Bisociative Knowledge Discovery Motivation Bisociation Types of Bisociation Bridging Concepts Bridging Graphs Bridging by Structural Similarity Other Types of Bisociation Bisociation Discovery Methods Future Directions Conclusions References Towards Creative Information Exploration Based on Koestler’s Concept of Bisociation Introduction Creativity What Is Creativity? Three Roads to Creativity Computational Creativity Koestler's Concept of Bisociation Elements of Bisociative Computational Creativity Towards a Formal Definition of Bisociation Related Work Discussion and Conclusion References From Information Networks to Bisociative Information Networks Introduction Different Categories of Information Network Properties of Information Units Properties of Relations Prominent Types of Information Networks Ontologies Semantic Networks Topic Maps Weighted Networks BisoNets: Bisociative Information Networks Summary Patterns of Bisociation in BisoNets Bridging Concept Bridging Graphs Bridging by Graph Similarity Conclusion References Part II: Representation and Network Creation Network Creation: Overview References Selecting the Links in BisoNets Generated from Document Collections Introduction Reminder: Bisociation and BisoNets BisoNet Generation Data Access and Pre-processing Creating Nodes Linking Nodes: Different Metrics Cosine and Tanimoto Measures The Bison Measure The Probabilistic Measure Benchmarks The Swanson Benchmark The Biology and Music Benchmark Conclusion References Bridging Concept Identification for Constructing Information Networks from Text Documents Introduction Problem Description Document Acquisition and Preprocessing Document Acquisition Document Preprocessing Background Knowledge Candidate Concept Detection Distance Measures between Vectors Identifying Bridging Concept Candidates for High Quality Network Entities Extraction Heuristics Description Frequency Based Heuristics Tf-idf Based Heuristics Similarity Based Heuristics Outlier Based Heuristics Baseline Heuristics Heuristics Evaluation Evaluation Procedure Migraine-Magnesium Dataset Comparison of the Heuristics Network Creation References Discovery of Novel Term Associations in a Document Collection Introduction Related Work The tpf–idf–tpu Model of Important Term Pair Associations Term Pair Frequency (tpf) and Inverse Document Frequency (idf) Term Pair Uncorrelation (tpu) Experiments Tpf–idf–tpu vs. tf–idf Sentence vs. Document-Level tpf–idf–tpu Methods Comparison of tpf–idf–tpu and tf–idf Using Annotated Test Set Conclusion References Cover Similarity Based Item Set Mining Introduction Frequent Item Set Mining Jaccard Item Sets The Eclat Algorithm The JIM Algorithm (Jaccard Item Set Mining) Other Similarity Measures Experiments Conclusions References Patterns and Logic for Reasoning with Networks Introduction The Biomine and ProbLog Frameworks Using Graphs: Biomine Using Logic: ProbLog Summary Inference and Reasoning Techniques Deduction: Reasoning about Node Tuples Abduction: Reasoning about Subgraphs Induction: Finding Patterns Combining Induction and Deduction Modifying the Knowledge Base Summary Using Probabilistic or Algebraic Labels The Probabilistic Model of Biomine and ProbLog Probabilistic Deduction Probabilistic Abduction and Top-k Instantiations Patterns and Probabilities Combining Induction and Deduction Modifying the Probabilistic Knowledge Base Beyond Probabilities Conclusions References Part III: Network Analysis Network Analysis: Overview References BiQL: A Query Language for Analyzing Information Networks Introduction Motivating Example Requirements Data Representation Basic Data Manipulation Illustrative Examples Related Work Knowledge Discovery Databases Conclusions References Review of BisoNet Abstraction Techniques Introduction Preference-Free Methods Relative Neighborhood Graph Node Centrality PageRank and HITS Birnbaum's Component Importance Graph Partitioning Hierarchical Clustering Edge Betweenness Frequent Subgraphs Preference-Dependent Methods Relevant Subgraph Extraction Detecting Interesting Nodes or Paths Personalized PageRank Exact Subgraph Search Similarity Subgraph Search Conclusion References Simplification of Networks by Edge Pruning Introduction Lossy Network Simplification Definitions Example Instances of the Framework Analysis of the Problem Multiplicativity of Ratio of Connectivity Kept A Bound on the Ratio of Connectivity Kept A Further Bound on the Ratio of Connectivity Kept Algorithms Naive Approach Brute Force Approach Path Simplification Combinational Approach Experiments Experimental Setup Results Related Work Conclusion References Network Compression by Node and Edge Mergers Introduction Problem Definition Weighted and Compressed Graphs Simple Weighted Graph Compression Generalized Weighted Graph Compression Optimal Superedge Weights and Mergers Bounds for Distances between Graphs A Bound on Distances between Nodes Related Work Algorithms Experiments Experimental Setup Results Conclusions References Finding Representative Nodes in Probabilistic Graphs Introduction Related Work Similarities in Probabilistic Graphs Clustering and Representatives in Graphs Experiments Test Setting Results Conclusions References (Missing) Concept Discovery in Heterogeneous Information Networks Introduction Bisociative Information Networks Concept Graphs Preliminaries Detection Application Results Conclusion and Future work References Node Similarities from Spreading Activation Introduction Related Work Spreading Activation Linear Standard Scenario Node Signatures Node Similarities Activation Similarity Signature Similarity Experiments Schools-Wikipedia Conclusion References Towards Discovery of Subgraph Bisociations Motivation Networks, Domains and Bisociations Knowledge Modeling Domains Bisociations Finding and Assessing Bisociations Domain Extraction Scoring Bisociation Candidates Complexity and Scalability Preliminary Evaluation Related Work Conclusion References Part IV: Exploration Exploration: Overview Introduction Contributions Conclusions References Data Exploration for Bisociative Knowledge Discovery: A Brief Overview of Tools and Evaluation Methods Introduction Bisociative Data Exploration Different Meanings of Exploration Definition of Bisociative Exploration Implications for User Interface Design Supporting Bisociative Data Exploration Tools for Data Exploration Evaluation of Knowledge Discovery Tools Evaluation Challenges Open Issues Benchmark Evaluation for Discovery Tools Conclusion and Future Work References On the Integration of Graph Exploration and Data Analysis: The Creative Exploration Toolkit Introduction State of the Art in Graph Interaction and Visualization The Creative Exploration Toolkit Network and Algorithm Providers Communication between CET and Other Tools The KNIME Information Mining Platform Wikipedia Evaluation Study Design Results of the Study Conclusion and Future Work References Bisociative Knowledge Discovery by Literature Outlier Detection Introduction Related Work in Literature Mining The Upgraded RaJoLink Knowledge Discovery Process Outlier Detection in the RaJoLink Knowledge Discovery Process Application of Outlier Detection in the Autism Literature Conclusions References Exploring the Power of Outliers for Cross-Domain Literature Mining Introduction Related Work Experimental Datasets Detecting Outlier Documents Classification Noise Filters for Outlier Detection Experimental Evaluation Conclusions References Bisociative Literature Mining by Ensemble Heuristics Introduction Problem Description Methodology for Bridging Concept Identification and Ranking Base Heuristics Ensemble Heuristic Evaluation of the Methodology Experimental Setting Results in the Migraine-Magnesium Dataset Results in Autism-Calcineurin Dataset The CrossBee System A Typical Use Case Other CrossBee Functionalities Discussion and Further Work References Part V: Applications and Evaluation Applications and Evaluation: Overview Introduction Contributions Lessons Learned The BISON Software for Applications Development Application Potential of the BISON Methodology Evaluation of the BISON Methodology and the Potential for Triggering Creativity The Future of Bisociative Reasoning and Cross-Context Data Mining References Biomine: A Network-Structured Resource of Biological Entities for Link Prediction Introduction Biomine Database Data Model Source Databases Edge Goodness in Biomine Link Goodness Measures Path and Neighbourhood Level Subgraph Level Graph Level Estimation of Link Significance Related Work Conclusion References Semantic Subgroup Discovery and Cross-Context Linking for Microarray Data Analysis Introduction Exploratory Gene Analytics SEGS: Search for Enriched Gene Sets SegMine: Contextualization of genes A Case Study Conclusions References Contrast Mining from Interesting Subgroups Introduction Related Work Contrast Mining from Interesting Subgroups Subgroup Discovery (Step 1) Construction of Contrast Classes (Step 2) Subgroup Discovery (Step 3) An Instance of Our Method: Gene Set Enrichment from Enriched Gene Sets Experiments Conclusion References Link and Node Prediction in Metabolic Networks with Probabilistic Logic Introduction The Probabilistic Logic Environment: ProbLog Method Metabolic Network Representation Models for Automatic Network Curation Experimental Setup Agnostic Noise Model Noise Model for Unreliable Predictions Conclusions References Modelling a Biological System: Network Creation by Triplet Extraction from Biological Literature Introduction Related Work Manually Constructed Petri Net Model of Plant Defence Response Automated Extraction of Plant Defence Response Model from Biological Literature Two Modelling Scenarios An Illustrative Example Crossing the Boundaries of Individual Readers Conclusion References Bisociative Exploration of Biological and Financial Literature Using Clustering Introduction The Bile Acid and Xenobiotic System Materials and Methods Results and Discussion Conclusions References Bisociative Discovery in Business Process Models Introduction Related Work Interfaces for Creative Music Discovery Projection of a Similarity Space User-Adaption during the Exploration Process The MusicGalaxy Visualization Bisociative Lens Distortions Orthogonal Similarity Measures Generalization to Domain Graphs Discussion Conclusions References Author Index 1. Towards Bisociative Knowledge Discovery Towards Bisociative Knowledge Discovery Motivation Bisociation Types of Bisociation Bridging Concepts Bridging Graphs Bridging by Structural Similarity Other Types of Bisociation Bisociation Discovery Methods Future Directions Conclusions References 2. Towards Creative Information Exploration Based on Koestlerâ•Žs Concept of Bisociation Towards Creative Information Exploration Based on Koestler’s Concept of Bisociation Introduction Creativity What Is Creativity? Three Roads to Creativity Computational Creativity Koestler's Concept of Bisociation Elements of Bisociative Computational Creativity Towards a Formal Definition of Bisociation Related Work Discussion and Conclusion References 3. From Information Networks to Bisociative Information Networks From Information Networks to Bisociative Information Networks Introduction Different Categories of Information Network Properties of Information Units Properties of Relations Prominent Types of Information Networks Ontologies Semantic Networks Topic Maps Weighted Networks BisoNets: Bisociative Information Networks Summary Patterns of Bisociation in BisoNets Bridging Concept Bridging Graphs Bridging by Graph Similarity Conclusion References 4. Network Creation: Overview Network Creation: Overview References 5. Selecting the Links in BisoNets Generated from Document Collections Selecting the Links in BisoNets Generated from Document Collections Introduction Reminder: Bisociation and BisoNets BisoNet Generation Data Access and Pre-processing Creating Nodes Linking Nodes: Different Metrics Cosine and Tanimoto Measures The Bison Measure The Probabilistic Measure Benchmarks The Swanson Benchmark The Biology and Music Benchmark Conclusion References 6. Bridging Concept Identification for Constructing Information Networks from Text Documents Bridging Concept Identification for Constructing Information Networks from Text Documents Introduction Problem Description Document Acquisition and Preprocessing Document Acquisition Document Preprocessing Background Knowledge Candidate Concept Detection Distance Measures between Vectors Identifying Bridging Concept Candidates for High Quality Network Entities Extraction Heuristics Description Frequency Based Heuristics Tf-idf Based Heuristics Similarity Based Heuristics Outlier Based Heuristics Baseline Heuristics Heuristics Evaluation Evaluation Procedure Migraine-Magnesium Dataset Comparison of the Heuristics Network Creation References 7. Discovery of Novel Term Associations in a Document Collection Discovery of Novel Term Associations in a Document Collection Introduction Related Work The tpf–idf–tpu Model of Important Term Pair Associations Term Pair Frequency (tpf) and Inverse Document Frequency (idf) Term Pair Uncorrelation (tpu) Experiments Tpf–idf–tpu vs. tf–idf Sentence vs. Document-Level tpf–idf–tpu Methods Comparison of tpf–idf–tpu and tf–idf Using Annotated Test Set Conclusion References 8. Cover Similarity Based Item Set Mining Cover Similarity Based Item Set Mining Introduction Frequent Item Set Mining Jaccard Item Sets The Eclat Algorithm The JIM Algorithm (Jaccard Item Set Mining) Other Similarity Measures Experiments Conclusions References 9. Patterns and Logic for Reasoning with Networks Patterns and Logic for Reasoning with Networks Introduction The Biomine and ProbLog Frameworks Using Graphs: Biomine Using Logic: ProbLog Summary Inference and Reasoning Techniques Deduction: Reasoning about Node Tuples Abduction: Reasoning about Subgraphs Induction: Finding Patterns Combining Induction and Deduction Modifying the Knowledge Base Summary Using Probabilistic or Algebraic Labels The Probabilistic Model of Biomine and ProbLog Probabilistic Deduction Probabilistic Abduction and Top-k Instantiations Patterns and Probabilities Combining Induction and Deduction Modifying the Probabilistic Knowledge Base Beyond Probabilities Conclusions References 10. Network Analysis: Overview Network Analysis: Overview References 11. BiQL: A Query Language for Analyzing Information Networks BiQL: A Query Language for Analyzing Information Networks Introduction Motivating Example Requirements Data Representation Basic Data Manipulation Illustrative Examples Related Work Knowledge Discovery Databases Conclusions References 12. Review of BisoNet Abstraction Techniques Review of BisoNet Abstraction Techniques Introduction Preference-Free Methods Relative Neighborhood Graph Node Centrality PageRank and HITS Birnbaum's Component Importance Graph Partitioning Hierarchical Clustering Edge Betweenness Frequent Subgraphs Preference-Dependent Methods Relevant Subgraph Extraction Detecting Interesting Nodes or Paths Personalized PageRank Exact Subgraph Search Similarity Subgraph Search Conclusion References 13. Simplification of Networks by Edge Pruning Simplification of Networks by Edge Pruning Introduction Lossy Network Simplification Definitions Example Instances of the Framework Analysis of the Problem Multiplicativity of Ratio of Connectivity Kept A Bound on the Ratio of Connectivity Kept A Further Bound on the Ratio of Connectivity Kept Algorithms Naive Approach Brute Force Approach Path Simplification Combinational Approach Experiments Experimental Setup Results Related Work Conclusion References 14. Network Compression by Node and Edge Mergers Network Compression by Node and Edge Mergers Introduction Problem Definition Weighted and Compressed Graphs Simple Weighted Graph Compression Generalized Weighted Graph Compression Optimal Superedge Weights and Mergers Bounds for Distances between Graphs A Bound on Distances between Nodes Related Work Algorithms Experiments Experimental Setup Results Conclusions References 15. Finding Representative Nodes in Probabilistic Graphs Finding Representative Nodes in Probabilistic Graphs Introduction Related Work Similarities in Probabilistic Graphs Clustering and Representatives in Graphs Experiments Test Setting Results Conclusions References 16. (Missing) Concept Discovery in Heterogeneous Information Networks (Missing) Concept Discovery in Heterogeneous Information Networks Introduction Bisociative Information Networks Concept Graphs Preliminaries Detection Application Results Conclusion and Future work References 17. Node Similarities from Spreading Activation Node Similarities from Spreading Activation Introduction Related Work Spreading Activation Linear Standard Scenario Node Signatures Node Similarities Activation Similarity Signature Similarity Experiments Schools-Wikipedia Conclusion 18. Towards Discovery of Subgraph Bisociations Towards Discovery of Subgraph Bisociations Motivation Networks, Domains and Bisociations Knowledge Modeling Domains Bisociations Finding and Assessing Bisociations Domain Extraction Scoring Bisociation Candidates Complexity and Scalability Preliminary Evaluation Related Work Conclusion References 19. Exploration: Overview Exploration: Overview Introduction Contributions Conclusions References 20. Data Exploration for Bisociative Knowledge Discovery: A Brief Overview of Tools and Evaluation Methods Data Exploration for Bisociative Knowledge Discovery: A Brief Overview of Tools and Evaluation Methods Introduction Bisociative Data Exploration Different Meanings of Exploration Definition of Bisociative Exploration Implications for User Interface Design Supporting Bisociative Data Exploration Tools for Data Exploration Evaluation of Knowledge Discovery Tools Evaluation Challenges Open Issues Benchmark Evaluation for Discovery Tools Conclusion and Future Work References 21. On the Integration of Graph Exploration and Data Analysis: The Creative Exploration Toolkit On the Integration of Graph Exploration and Data Analysis: The Creative Exploration Toolkit Introduction State of the Art in Graph Interaction and Visualization The Creative Exploration Toolkit Network and Algorithm Providers Communication between CET and Other Tools The KNIME Information Mining Platform Wikipedia Evaluation Study Design Results of the Study Conclusion and Future Work References 22. Bisociative Knowledge Discovery by Literature Outlier Detection Bisociative Knowledge Discovery by Literature Outlier Detection Introduction Related Work in Literature Mining The Upgraded RaJoLink Knowledge Discovery Process Outlier Detection in the RaJoLink Knowledge Discovery Process Application of Outlier Detection in the Autism Literature Conclusions References 23. Exploring the Power of Outliers for Cross-Domain Literature Mining Exploring the Power of Outliers for Cross-Domain Literature Mining Introduction Related Work Experimental Datasets Detecting Outlier Documents Classification Noise Filters for Outlier Detection Experimental Evaluation Conclusions References 24. Bisociative Literature Mining by Ensemble Heuristics Bisociative Literature Mining by Ensemble Heuristics Introduction Problem Description Methodology for Bridging Concept Identification and Ranking Base Heuristics Ensemble Heuristic Evaluation of the Methodology Experimental Setting Results in the Migraine-Magnesium Dataset Results in Autism-Calcineurin Dataset The CrossBee System A Typical Use Case Other CrossBee Functionalities Discussion and Further Work References 25. Applications and Evaluation: Overview Applications and Evaluation: Overview Introduction Contributions Lessons Learned The BISON Software for Applications Development Application Potential of the BISON Methodology Evaluation of the BISON Methodology and the Potential for Triggering Creativity The Future of Bisociative Reasoning and Cross-Context Data Mining References 26. Biomine: A Network-Structured Resource of Biological Entities for Link Prediction Biomine: A Network-Structured Resource of Biological Entities for Link Prediction Introduction Biomine Database Data Model Source Databases Edge Goodness in Biomine Link Goodness Measures Path and Neighbourhood Level Subgraph Level Graph Level Estimation of Link Significance Related Work Conclusion References 27. Semantic Subgroup Discovery and Cross-Context Linking for Microarray Data Analysis Semantic Subgroup Discovery and Cross-Context Linking for Microarray Data Analysis Introduction Exploratory Gene Analytics SEGS: Search for Enriched Gene Sets SegMine: Contextualization of genes A Case Study Conclusions References 28. Contrast Mining from Interesting Subgroups Contrast Mining from Interesting Subgroups Introduction Related Work Contrast Mining from Interesting Subgroups Subgroup Discovery (Step 1) Construction of Contrast Classes (Step 2) Subgroup Discovery (Step 3) An Instance of Our Method: Gene Set Enrichment from Enriched Gene Sets Experiments Conclusion References 29. Link and Node Prediction in Metabolic Networks with Probabilistic Logic Link and Node Prediction in Metabolic Networks with Probabilistic Logic Introduction The Probabilistic Logic Environment: ProbLog Method Metabolic Network Representation Models for Automatic Network Curation Experimental Setup Agnostic Noise Model Noise Model for Unreliable Predictions Conclusions References 30. Modelling a Biological System: Network Creation by Triplet Extraction from Biological Literature Modelling a Biological System: Network Creation by Triplet Extraction from Biological Literature Introduction Related Work Manually Constructed Petri Net Model of Plant Defence Response Automated Extraction of Plant Defence Response Model from Biological Literature Two Modelling Scenarios An Illustrative Example Crossing the Boundaries of Individual Readers Conclusion References 31. Bisociative Exploration of Biological and Financial Literature Using Clustering Bisociative Exploration of Biological and Financial Literature Using Clustering Introduction The Bile Acid and Xenobiotic System Materials and Methods Results and Discussion Conclusions References 32. Bisociative Discovery in Business Process Models Bisociative Discovery in Business Process Models Introduction Tools Used for Pre-processing Data Fuzzy Grammars Fuzzy Formal Concept Analysis Process Data Bisociative Knowledge Discovery in Business Processes Illustrative Example Business Process Example - Definition of Domains Bisociations Summary References 33. Bisociative Music Discovery and Recommendation Bisociative Music Discovery and Recommendation Introduction Related Work Interfaces for Creative Music Discovery Projection of a Similarity Space User-Adaption during the Exploration Process The MusicGalaxy Visualization Bisociative Lens Distortions Orthogonal Similarity Measures Generalization to Domain Graphs Discussion Conclusions References Back matter Title Foreword Table of Contents Part I: Bisociation Towards Bisociative Knowledge Discovery Motivation Bisociation Types of Bisociation Bridging Concepts Bridging Graphs Bridging by Structural Similarity Other Types of Bisociation Bisociation Discovery Methods Future Directions Conclusions References Towards Creative Information Exploration Based on Koestler’s Concept of Bisociation Introduction Creativity What Is Creativity? Three Roads to Creativity Computational Creativity Koestler's Concept of Bisociation Elements of Bisociative Computational Creativity Towards a Formal Definition of Bisociation Related Work Discussion and Conclusion References From Information Networks to Bisociative Information Networks Introduction Different Categories of Information Network Properties of Information Units Properties of Relations Prominent Types of Information Networks Ontologies Semantic Networks Topic Maps Weighted Networks BisoNets: Bisociative Information Networks Summary Patterns of Bisociation in BisoNets Bridging Concept Bridging Graphs Bridging by Graph Similarity Conclusion References Part II: Representation and Network Creation Network Creation: Overview References Selecting the Links in BisoNets Generated from Document Collections Introduction Reminder: Bisociation and BisoNets BisoNet Generation Data Access and Pre-processing Creating Nodes Linking Nodes: Different Metrics Cosine and Tanimoto Measures The Bison Measure The Probabilistic Measure Benchmarks The Swanson Benchmark The Biology and Music Benchmark Conclusion References Bridging Concept Identification for Constructing Information Networks from Text Documents Introduction Problem Description Document Acquisition and Preprocessing Document Acquisition Document Preprocessing Background Knowledge Candidate Concept Detection Distance Measures between Vectors Identifying Bridging Concept Candidates for High Quality Network Entities Extraction Heuristics Description Frequency Based Heuristics Tf-idf Based Heuristics Similarity Based Heuristics Outlier Based Heuristics Baseline Heuristics Heuristics Evaluation Evaluation Procedure Migraine-Magnesium Dataset Comparison of the Heuristics Network Creation References Discovery of Novel Term Associations in a Document Collection Introduction Related Work The tpf–idf–tpu Model of Important Term Pair Associations Term Pair Frequency (tpf) and Inverse Document Frequency (idf) Term Pair Uncorrelation (tpu) Experiments Tpf–idf–tpu vs. tf–idf Sentence vs. Document-Level tpf–idf–tpu Methods Comparison of tpf–idf–tpu and tf–idf Using Annotated Test Set Conclusion References Cover Similarity Based Item Set Mining Introduction Frequent Item Set Mining Jaccard Item Sets The Eclat Algorithm The JIM Algorithm (Jaccard Item Set Mining) Other Similarity Measures Experiments Conclusions References Patterns and Logic for Reasoning with Networks Introduction The Biomine and ProbLog Frameworks Using Graphs: Biomine Using Logic: ProbLog Summary Inference and Reasoning Techniques Deduction: Reasoning about Node Tuples Abduction: Reasoning about Subgraphs Induction: Finding Patterns Combining Induction and Deduction Modifying the Knowledge Base Summary Using Probabilistic or Algebraic Labels The Probabilistic Model of Biomine and ProbLog Probabilistic Deduction Probabilistic Abduction and Top-k Instantiations Patterns and Probabilities Combining Induction and Deduction Modifying the Probabilistic Knowledge Base Beyond Probabilities Conclusions References Part III: Network Analysis Network Analysis: Overview References BiQL: A Query Language for Analyzing Information Networks Introduction Motivating Example Requirements Data Representation Basic Data Manipulation Illustrative Examples Related Work Knowledge Discovery Databases Conclusions References Review of BisoNet Abstraction Techniques Introduction Preference-Free Methods Relative Neighborhood Graph Node Centrality PageRank and HITS Birnbaum's Component Importance Graph Partitioning Hierarchical Clustering Edge Betweenness Frequent Subgraphs Preference-Dependent Methods Relevant Subgraph Extraction Detecting Interesting Nodes or Paths Personalized PageRank Exact Subgraph Search Similarity Subgraph Search Conclusion References Simplification of Networks by Edge Pruning Introduction Lossy Network Simplification Definitions Example Instances of the Framework Analysis of the Problem Multiplicativity of Ratio of Connectivity Kept A Bound on the Ratio of Connectivity Kept A Further Bound on the Ratio of Connectivity Kept Algorithms Naive Approach Brute Force Approach Path Simplification Combinational Approach Experiments Experimental Setup Results Related Work Conclusion References Network Compression by Node and Edge Mergers Introduction Problem Definition Weighted and Compressed Graphs Simple Weighted Graph Compression Generalized Weighted Graph Compression Optimal Superedge Weights and Mergers Bounds for Distances between Graphs A Bound on Distances between Nodes Related Work Algorithms Experiments Experimental Setup Results Conclusions References Finding Representative Nodes in Probabilistic Graphs Introduction Related Work Similarities in Probabilistic Graphs Clustering and Representatives in Graphs Experiments Test Setting Results Conclusions References (Missing) Concept Discovery in Heterogeneous Information Networks Introduction Bisociative Information Networks Concept Graphs Preliminaries Detection Application Results Conclusion and Future work References Node Similarities from Spreading Activation Introduction Related Work Spreading Activation Linear Standard Scenario Node Signatures Node Similarities Activation Similarity Signature Similarity Experiments Schools-Wikipedia Conclusion References Towards Discovery of Subgraph Bisociations Motivation Networks, Domains and Bisociations Knowledge Modeling Domains Bisociations Finding and Assessing Bisociations Domain Extraction Scoring Bisociation Candidates Complexity and Scalability Preliminary Evaluation Related Work Conclusion References Part IV: Exploration Exploration: Overview Introduction Contributions Conclusions References Data Exploration for Bisociative Knowledge Discovery: A Brief Overview of Tools and Evaluation Methods Introduction Bisociative Data Exploration Different Meanings of Exploration Definition of Bisociative Exploration Implications for User Interface Design Supporting Bisociative Data Exploration Tools for Data Exploration Evaluation of Knowledge Discovery Tools Evaluation Challenges Open Issues Benchmark Evaluation for Discovery Tools Conclusion and Future Work References On the Integration of Graph Exploration and Data Analysis: The Creative Exploration Toolkit Introduction State of the Art in Graph Interaction and Visualization The Creative Exploration Toolkit Network and Algorithm Providers Communication between CET and Other Tools The KNIME Information Mining Platform Wikipedia Evaluation Study Design Results of the Study Conclusion and Future Work References Bisociative Knowledge Discovery by Literature Outlier Detection Introduction Related Work in Literature Mining The Upgraded RaJoLink Knowledge Discovery Process Outlier Detection in the RaJoLink Knowledge Discovery Process Application of Outlier Detection in the Autism Literature Conclusions References Exploring the Power of Outliers for Cross-Domain Literature Mining Introduction Related Work Experimental Datasets Detecting Outlier Documents Classification Noise Filters for Outlier Detection Experimental Evaluation Conclusions References Bisociative Literature Mining by Ensemble Heuristics Introduction Problem Description Methodology for Bridging Concept Identification and Ranking Base Heuristics Ensemble Heuristic Evaluation of the Methodology Experimental Setting Results in the Migraine-Magnesium Dataset Results in Autism-Calcineurin Dataset The CrossBee System A Typical Use Case Other CrossBee Functionalities Discussion and Further Work References Part V: Applications and Evaluation Applications and Evaluation: Overview Introduction Contributions Lessons Learned The BISON Software for Applications Development Application Potential of the BISON Methodology Evaluation of the BISON Methodology and the Potential for Triggering Creativity The Future of Bisociative Reasoning and Cross-Context Data Mining References Biomine: A Network-Structured Resource of Biological Entities for Link Prediction Introduction Biomine Database Data Model Source Databases Edge Goodness in Biomine Link Goodness Measures Path and Neighbourhood Level Subgraph Level Graph Level Estimation of Link Significance Related Work Conclusion References Semantic Subgroup Discovery and Cross-Context Linking for Microarray Data Analysis Introduction Exploratory Gene Analytics SEGS: Search for Enriched Gene Sets SegMine: Contextualization of genes A Case Study Conclusions References Contrast Mining from Interesting Subgroups Introduction Related Work Contrast Mining from Interesting Subgroups Subgroup Discovery (Step 1) Construction of Contrast Classes (Step 2) Subgroup Discovery (Step 3) An Instance of Our Method: Gene Set Enrichment from Enriched Gene Sets Experiments Conclusion References Link and Node Prediction in Metabolic Networks with Probabilistic Logic Introduction The Probabilistic Logic Environment: ProbLog Method Metabolic Network Representation Models for Automatic Network Curation Experimental Setup Agnostic Noise Model Noise Model for Unreliable Predictions Conclusions References Modelling a Biological System: Network Creation by Triplet Extraction from Biological Literature Introduction Related Work Manually Constructed Petri Net Model of Plant Defence Response Automated Extraction of Plant Defence Response Model from Biological Literature Two Modelling Scenarios An Illustrative Example Crossing the Boundaries of Individual Readers Conclusion References Bisociative Exploration of Biological and Financial Literature Using Clustering Introduction The Bile Acid and Xenobiotic System Materials and Methods Results and Discussion Conclusions References Bisociative Discovery in Business Process Models Introduction Related Work Interfaces for Creative Music Discovery Projection of a Similarity Space User-Adaption during the Exploration Process The MusicGalaxy Visualization Bisociative Lens Distortions Orthogonal Similarity Measures Generalization to Domain Graphs Discussion Conclusions References Author Index