Intro
Preface
Organization
Contents
Research Track
PCSG: Pattern-Coverage Snippet Generation for RDF Datasets
1 Introduction
2 Related Work
3 Snippet Generation: A Basic Approach
3.1 Problem Formulation
3.2 Algorithm Basic
4 Snippet Generation: Extended Approaches
4.1 Extension to Disconnectivity: Algorithm PCSG
4.2 Extension to High Heterogeneity: Algorithm PCSG-
4.3 Extension to Query Relevance: Algorithm QPCSG(
)
5 Evaluation of PCSG(
)
5.1 RDF Datasets
5.2 Participating Methods
5.3 Experiment 1: Coverage of Schema

5.4 Experiment 2: User Preference
5.5 Experiment 3: Space Saving and Run-Time
6 Evaluation of QPCSG(
)
6.1 Queries and RDF Datasets
6.2 Participating Methods
6.3 Experiment 4: Coverage of Query and Schema
6.4 Experiment 5: Space Saving and Run-Time
7 Empirical Comparison with Summary
8 Conclusion and Future Work
References
A Source-to-Target Constraint Rewriting for Direct Mapping
1 Introduction
2 Preliminaries
3 Constraint Rewriting: Definition and Properties
4 The Constraint Rewriting
4.1 Datalog Predicates
4.2 The Constraint Rewriting Rules

5 Properties of the Constraint Rewriting
6 Discussion
7 Conclusion
References
Learning to Predict the Departure Dynamics of Wikidata Editors
1 Introduction
2 Related Work
3 Proposed Approach
3.1 Statistical Features
3.2 Pattern-Based Features
3.3 DeepFM as the Classification Model
4 Wikidata Dataset
4.1 Exploratory Analysis of the Dataset
5 Experimental Setup
5.1 Dataset
5.2 Compared Methods
5.3 Evaluation Metrics
6 Results
6.1 Comparison with the Set of Different Methods
6.2 Analysis of Statistical Features
7 Conclusions
References

Towards Neural Schema Alignment for OpenStreetMap and Knowledge Graphs
1 Introduction
2 Problem Statement
3 Neural Class Alignment Approach
3.1 Auxiliary Neural Classification Model
3.2 Tag-to-Class Alignment
3.3 Illustrative Example
4 Evaluation Setup
4.1 Datasets
4.2 Ground Truth Creation
4.3 Baselines
4.4 Metrics
5 Evaluation
5.1 Tag-to-Class Alignment Performance
5.2 Influence of the Shared Latent Space
5.3 Confidence Threshold Tuning
5.4 Alignment Impact
6 Related Work
7 Conclusion
References

Improving Inductive Link Prediction Using Hyper-relational Facts
1 Introduction
2 Background
2.1 Statements: Triples Plus Qualifiers
2.2 Expressiveness
3 Inductive Link Prediction
4 Approach
4.1 Encoders
4.2 Decoder
4.3 Training
5 Datasets
5.1 Fully-Inductive Setting
5.2 Semi-inductive Setting
5.3 Overview
6 Experiments
6.1 Experimental Setup
6.2 Fully-Inductive Setting
6.3 Semi-inductive Setting
6.4 Qualitative Analysis
7 Related Work
8 Conclusion
A Training
B Hyperparameter Ranges
C Infrastructure and Parameters