From Data to Models and Back: Revised Selected Papers of the 9th International Symposium, DataMod 2020, Virtual Event, October 20, 2020

This book constitutes the refereed proceedings of the 9th International Symposium on From Data Models and Back, DataMod 2020, held virtually, in October 2020. The 11 full papers and 3 short papers presented in this book were selected from 19 submissions. The papers are grouped in these topical sections: machine learning; simulation-based approaches, and data mining and processing related approaches. Preface 6 Organization 8 Towards AI-driven Data Analysis and Fabrication (Abstract of Invited Talk) 9 Contents 10 Machine Learning 12 Synthesis and Pruning as a Dynamic Compression Strategy for Efficient Deep Neural Networks 13 Abstract 13 1 Introduction 13 2 Related Work 14 3 Method 16 3.1 Approaches 16 3.2 Procedure 17 4 Problem Dataset 19 5 Results 19 6 Further Considerations 22 6.1 Residual Networks 22 6.2 Sub-networks 23 6.3 False Starts 23 6.4 Strategic Targets 23 6.5 Redundancy 24 6.6 Hybrid Scheduling 24 7 Conclusions and Future Work 24 References 26 Exploring Graph-Based Neural Networks for Automatic Brain Tumor Segmentation 28 1 Introduction 28 2 Related Work 30 2.1 Convolutional Neural Networks 30 2.2 Graph Neural Networks 31 2.3 Explanation of Deep Learning Models 32 3 Methods 32 3.1 Imaging Data 32 3.2 Data Preprocessing 33 3.3 Graph Construction 33 3.4 GNN Details 35 3.5 Training and Evaluation Metrics 35 3.6 Baseline Method 36 3.7 Model Interpretation 37 4 Results 37 4.1 Supervoxel Generation Affects Achievable Accuracy 37 4.2 Brain Tumor Segmentation Performance of Different GNN Models 38 4.3 Performance and Runtime Results for Varying Neighborhood Sizes 40 4.4 Comparison of GNN Model with Other Recent Models 40 4.5 Explaining GNN Predictions Using SHAP 41 5 Discussion 43 References 45 STDI-Net: Spatial-Temporal Network with Dynamic Interval Mapping for Bike Sharing Demand Prediction 48 1 Introduction 48 2 Related Work 51 3 Preliminaries 51 4 Proposed Spatial-Temporal Dynamic Interval Network 52 4.1 Spatial Module 53 4.2 Temporal Module 54 4.3 Dynamic Interval Module 55 4.4 Implementation Details 56 5 Experiment 57 5.1 Dataset 57 5.2 Evaluation Metric 57 5.3 Baselines 57 5.4 Comparison with Baselines 58 5.5 Comparison with Modules Combinations 59 5.6 Comparison with Variants of Our Model 60 5.7 Influence of Sequence Length and Number of ResUnits 61 6 Conclusion and Discussion 62 References 62 Simulation-Based Approaches 64 A Simulation-Based Approach for the Behavioural Analysis of Cancer Pathways 65 1 Introduction 65 2 Data Analysis and Queries Development 67 3 Simulation Modelling and DES 69 4 Model Validation 75 5 Simulation Results 76 6 Conclusion 78 References 78 Discovering the Impact of Notifications on Social Network Addiction 80 1 Introduction 80 2 The Internet Addiction 81 2.1 The Network Communication Model 82 2.2 The Mathematical Model of Dopamine System 83 3 Dataset 86 4 Simulations 87 5 Results 88 6 Conclusions 92 References 93 A Simulation Study on Demand Disruptions and Limited Resources for Healthcare Provision 95 1 Introduction 95 2 Southern Brazilian Hospital Settings 96 2.1 A&E Triage System 97 2.2 Usual Pre-pandemic A&E Patient Care Process 98 3 Applied Modelling and Simulation in the A&E 100 3.1 Input Data Modelling for the Pre-pandemic Simulation 102 4 Pre-pandemic and Post-pandemic Scenarios Simulation 105 4.1 Pre-pandemic Scenario Simulation 106 4.2 A&E Disruption Scenarios During and Post-pandemic 108 5 Conclusion 109 References 110 A Formal Model for Emulating the Generation of Human Knowledge in Semantic Memory 112 1 Introduction 112 2 Cognitive Models for Information Transfer 115 3 Real-Time Maude Models of STM and Semantic Memory 116 3.1 Facts, Questions and Goals 117 3.2 Modelling Basic Information Items and Goals 118 3.3 Modelling Explicit Attention and Goal Achievements 119 3.4 STM—Short-Term Memory 120 3.5 Semantic Memory 121 4 Modelling Memory Processes 122 4.1 Perception 122 4.2 Maintenance Rehearsal 124 4.3 Elaborative Rehearsal 125 4.4 Transfer from STM to Semantic Memory 126 5 In Silico Experiments 126 5.1 Rote Learning 127 5.2 Effective Learning 127 6 Conclusion and Future Work 128 References 129 Analysis of COVID-19 Data with PRISM: Parameter Estimation and SIR Modelling 131 1 Introduction 131 2 SIR Epidemic Models and COVID-19 132 3 Parameter Estimation 134 4 Translation into CTMC and Analysis with PRISM 137 5 Conclusions 139 References 140 A Formal Model for the Simulation and Analysis of Early Biofilm Formation 142 1 Introduction 142 1.1 Real-Time Maude 145 2 Biofilm Formation Model 146 2.1 Real-Time Maude Configuration and Rewrite Rules 147 2.2 Creation of the Initial Population 148 2.3 Biofilm Formation 148 2.4 Cell Reproduction 149 2.5 Cell Death 150 2.6 Time Increment 151 2.7 Intervention to Prevent Biofilm Formation 152 3 In Silico Experiments and Formal Analysis 153 3.1 Pseudomonas Aeruginosa 154 3.2 Simulation 154 3.3 Formal Analysis Using Model Checking 156 4 Conclusion and Future Work 156 References 157 Data Mining and Processing Related Approaches 160 Query Rewriting on Path Views Without Integrity Constraints 161 1 Introduction 161 2 Related Work 163 3 Preliminaries 164 4 Defining Smart Plans 166 5 Characterizing Smart Plans 168 5.1 Web Service Functions 168 5.2 Preliminary Definitions 168 5.3 Characterising Smart Plans 170 6 Generating Smart Plans 170 6.1 Minimal Smart Plans 171 6.2 Bounding and Generating the Weakly Smart Plans 171 6.3 Generating the Weakly Smart Plans 172 7 Experiments 175 7.1 Synthetic Functions 175 7.2 Real-World Web Services 176 8 Conclusion 178 References 178 Evaluating Trace Encoding Methods in Process Mining 180 1 Introduction 180 2 Related Work 181 3 Encoding Methods 182 3.1 Trace Replay and Alignment 182 3.2 Word Embeddings 184 3.3 Graph Embeddings 185 4 Materials and Methods 185 4.1 Event Logs 185 4.2 Trace Encoding 186 4.3 Feature Vector Measures and Classification Algorithm 187 5 Results and Discussion 187 5.1 Accuracy Performance 187 5.2 Time Usage 189 5.3 Encoding Representativeness 190 5.4 Encoding Ranking 193 6 Conclusion 194 References 194 Semantic Annotations in Clinical Guidelines 196 1 Introduction 196 2 Related Work 197 3 The Framework 198 4 Semantic Annotations 199 4.1 A Hierarchy of Concepts 199 4.2 Relationship Between Classes 200 4.3 Named-Entity Recognition 201 5 Learning 202 5.1 Syntactic Analysis of Guideline Sentences 202 5.2 Semantic Analysis of Guideline Sentences 203 5.3 Features 204 5.4 Learning Algorithm 205 5.5 Long Short-Term Memory for NER 205 6 Evaluation 207 6.1 Dataset Analysis 207 6.2 Experiments 208 7 Conclusion 209 References 210 Deriving Performance Measures of Workflow in Radiation Therapy from Real-Time Data 212 1 Introduction 212 2 Methods 214 2.1 Standardised Model of the RTP Workflow 214 2.2 Real-Time Tracking and Display of RTP Workflow 216 2.3 Performance Measures of RTP Workflow 216 3 Results 216 3.1 Real-Time Tracking and Display of Workflow 217 3.2 Performance Measures 217 3.3 Compliance in Recording Task Completion 218 3.4 Elapsed Time to Task Completion 218 3.5 On-Time Performance Relative to Ideal Timeline 219 3.6 Elapsed Time Between Tasks 219 4 Discussion 220 References 221 Handshape Classification in a Reverse Dictionary of Sign Languages for the Deaf 223 1 Introduction 223 2 Related Work 224 2.1 Sign Language Dictionaries 224 2.2 Sign Language Fingerspelling Recognition 224 3 System Design and Architecture 225 3.1 Datasets 226 3.2 Implementation Details 227 4 System Functionality 227 4.1 Search by Handshapes 228 4.2 Search by Fingerspelling 229 4.3 Other Search Methods 229 5 Evaluation 230 6 Conclusion and Future Work 230 References 231 Author Index 233