Fast Neuro-Symbolic Approaches for Class Expression Learning / N'Dah Jean Kouagou ; 1. Reviewer Prof. Dr. Axel-Cyrille Ngonga Ngomo (Heinz Nixdorf Institute, Department of Computer Science Paderborn University), 2. Reviewer Prof. Dr. Paul Groth (Informatics Institute University of Amsterdam) ; Supervisor Prof. Dr. Axel-Cyrille Ngonga Ngomo. Paderborn, 2025
Content
- Titlepage
- 1 Introduction
- 1.1 Motivation
- 1.2 Research Questions and Contributions
- 1.2.1 Concept Length Prediction
- 1.2.2 Neural Class Expression Synthesis
- 1.2.3 Neural Class Expression Synthesis in
- 1.2.4 Robust Class Expression Synthesis via Iterative Sampling
- 1.3 Thesis Outline
- 2 Background
- 2.1 Machine Learning
- 2.1.1 Overview
- 2.1.2 Neural Network Architectures
- 2.1.3 Permutation-Invariant Neural Network Architectures for Set Inputs
- 2.1.4 Expected and Empirical Risk Minimization
- 2.1.5 Loss Functions in Machine Learning
- 2.1.6 Optimization Algorithms for Machine Learning
- 2.2 Description Logics
- 2.3 Knowledge Base in Description Logics
- 2.4 Refinement Operators
- 2.5 Class Expression Learning
- 2.6 Knowledge Graph Embeddings
- 3 Related Work
- 3.1 Inductive Logic Programming
- 3.1.1 First Order Inductive Learner
- 3.1.2 First Order Combined Learner
- 3.1.3 Clausal Discovery Engine
- 3.2 State of the Art in Class Expression Learning
- 3.2.1 Description Logic First Order Inductive Learner
- 3.2.2 Description Logic First Order Combined Learner
- 3.2.3 OWL Class Expression Learner
- 3.2.4 Tree Learner
- 3.2.5 Class Expression Learner for Ontology Engineering
- 3.2.6 Drill
- 3.2.7 Efficient Concept Induction from Instances
- 3.2.8 Evolutionary Concept Learning
- 4 Concept Learner with Integrated Length Prediction
- 4.1 Methodology
- 4.1.1 Learning Problem
- 4.1.2 Overview of the Approach
- 4.1.3 Concept Length Prediction
- 4.1.4 Express Refinement
- 4.2 Experiments
- 4.3 Conclusion
- 5 Neural Class Expression Synthesis
- 6 Neural Class Expression Synthesis in
- 7 Robust Class Expression Synthesis via Iterative Sampling
- 8 Conclusion and Future Work
- 8.1 Summary
- 8.1.1 Concept Learner with Integrated Length Prediction
- 8.1.2 Neural Class Expression Synthesis
- 8.1.3 Neural Class Expression Synthesis in
- 8.1.4 Robust Class Expression Synthesis
- 8.2 Future Work
- Bibliography