Readings

Textbooks

Main course book (the whole book is mandatory reading):

• Hogan, A. et al. (2021). Knowledge Graphs. Springer. Synthesis Lectures on Data, Semantics, and Knowledge 22, 1–237, DOI: 10.2200/S01125ED1V01Y202109DSK022, Springer. https://kgbook.org/

Supplementary books (not mandatory):

• Dean Allemang, James Hendler & Fabien Gandon (2020). Semantic Web for the Working Ontologist, Effective Modeling for Linked Data, RDFS and OWL (Third Edition). ISBN: 9781450376143, PDF ISBN: 9781450376150, Hardcover ISBN: 9781450376174, DOI: 10.1145/3382097.
• Andreas Blumauer and Helmut Nagy (2020). The Knowledge Graph Cookbook - Recipes that Work. mono/monochrom. ISBN-10: ‎3902796707, ISBN-13: 978-3902796707.

Other materials

In addition, the materials listed below for each lecture are either mandatory or suggested reading. More materials will be added to each lecture in the coming weeks.

The labs, lectures and lectures notes are also part of the curriculum.

Make sure you download the electronic resources to your own computer in good time before the exam. This is your own responsibility. That way you are safe if a site becomes unavailable or somehow damaged the last few days before the exam.

Note: to download some of the papers, you may need to be inside UiB's network. Either use a computer directly on the UiB network or connect to your UiB account through VPN.

Lectures

Below are the mandatory and suggested readings for each lecture. All the textbook chapters in Hogan et al. ("Knowledge Graphs") are mandatory, whereas the chapters in Allemang, Hendler & Gandon ("Semantic Web") are suggested.

Session 1: Introduction to KGs

Themes:

• Introduction to Knowledge Graphs
• Organisation of the course

Mandatory readings:

• Chapter 1 Introduction, section 2.1 Models, and Appendix A Background in Hogan et al.
• Tim Berners-Lee talks about the semantic web
• RDFlib 7.1.3 documentation, the following pages:
  • The main page
  • Getting started with RDFLib
  • Loading and saving RDF
  • Creating RDF triples
  • Navigating Graphs
  • Utilities and convenience functions
  • RDF terms in rdflib
  • Namespaces and Bindings
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• Chapters 1-3 in Allemang, Hendler & Gandon (3rd edition)
• Wikidata (https://www.wikidata.org/)

Session 2: Querying and updating KGs (SPARQL)

Themes:

• SPARQL queries
• SPARQL Update
• Programming SPARQL and SPARQL Update in Python

Mandatory readings:

• Section 2.2 Queries in Hogan et al.
• The SPARQL query language — GraphDB 10.8 documentation
• rdflib 7.1.3 materials: Querying with SPARQL
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• Chapter 6 in Allemang, Hendler & Gandon (3rd edition)3.12 Session 12: KGs and LLMs
• SPARQL 1.1 Query Language
• SPARQL 1.1 Update Language
• SPARQL 1.1 Cheat Sheet
• SPARQL Expressions and Functions

Session 3: Creating KGs

Themes:

• Extracting KGs from text
• Extracting from marked-up sources
• Extracting from SQL databases and JSON

Mandatory readings:

• Chapter 6 Creation and Enrichment, sections 6.1-6.4, in Hogan et al.
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• DBpedia Spotlight
• Virtualization, GraphDB 10.0 documentation
• JSON for Linking Data

Session 4: Validating KGs

Themes:

• Validating KG schemas (SHACL)
• Semantic KG schemas/vocabularies (RDFS)

Mandatory readings:

• Section 3.1 Schema in Hogan et al.
• Sections 5.1, 5.3, 5.5, 5.6.1, and 5.6.3 in Gayo, J.E. et al. Validating RDF.
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• SHACL
  • Interactive, online SHACL Playground
  • pySHACL - A Python validator for SHACL at PyPi.org (after installation, go straight to "Python Module Use".)
  • Shapes Constraint Language (SHACL) (Editor's Draft)
• RDFS
  • W3C's RDF 1.1 Semantics (the axioms and entailments in sections 8 and 9 are most important, and we will go through the most important ones in the lecture)
  • OWL-RL adds inference capability on top of RDFLib. To use it, copy the owlrl folder into your project folder, next to your Python files, and import it with import owlrl.
  • OWL-RL documentation (most likely more detailed than you will need)

Session 5: Advanced KGs

Themes:

• More about RDF, e.g.,
  • identity
  • blank nodes
  • reification
  • higher-arity graphs

Mandatory readings:

• Sections 3.2 Identity and 3.3 Context in Hogan et al.
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• W3C's RDF 1.1 Concepts and Abstract Syntax
• RDF Grapher for drawing RDF graphs
• RDF Visualizer for drawing RDF graphs

Session 6: Ontologies

Themes:

• More powerful vocabularies/ontologies (OWL)
• Creating ontologies

Mandatory readings:

• Sections 4.1 Ontologies and 6.3 Schema/ontology creation in Hogan et al.
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• OWL 2 Primer, sections 2-6 (advanced: 9-10) (show: Turtle)
• WebVOWL interactive OWL visualisation tool
• Selected vocabularies:
  • Friend of a Friend (FOAF) (if necessary follow the link to the 2004 version)
  • Time ontology in OWL (time, OWL-time)
  • Dublin Core (DC)
  • SKOS - Simple Knowledge Organization System Home Page
  • Provenance Interchange (PROV)
• Linked Open Vocabularies (LOV, https://lov.linkeddata.es/dataset/lov/)

Session 7: Reasoning

Themes:

• More about semantic KG schemas (RDFS)
• Description logic
• OWL-DL

Mandatory readings:

• Section 4.2 Rules + DL in Hogan et al.
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• W3C's RDF 1.1 Semantics (the axioms and entailments in sections 8 and 9 are most important)
• W3C OWL 2 Overview
• W3C OWL 2 Primer
• W3C OWL 2 Quick Reference Guide (2nd Edition)

Session 8: KG Analytics

Themes:

• Graph analytics
  • graph metrics
  • directed vector-labelled graphs
  • analysis frameworks and techniques
• Symbolic learning
  • rule, axiom, and hypothesis mining

Mandatory readings:

• Sections 5.1 Graph Analytics and 5.4 Symbolic Learning in Hogan et al.
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Useful materials:

• NetworkX - Network analysis in Python

Session 9: KGs in Practice (Guest Lecture)

Guest lecture by Sindre Asplem, Capgemini.

Mandatory readings:

• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Session 10: KG Embeddings

Themes:

• Semantic embedding spaces
• KG embedding techniques
• Graph neural networks

Mandatory readings:

• Sections 5.2 Knowledge Graph Embeddings and 5.3 Graph neural networks in Hogan et al.
  • In Section 5.2.1, we focus on the Translational Models. The other models are cursory reading.
• Towards DataScience introduction: Introduction to Knowledge Graph Embeddings ( PDF)
• The slides from the lecture (available under Files/Slides in http://mitt.uib.no).

Supplementary readings:

• Towards DataScience introductions:
  • Introduction to Machine Learning for Beginners ( PDF)
  • Introduction to Word Embeddings and word2vec ( PDF)
• Mikolov et al’s original word2vec paper
• Bordes et al’s original TransE paper
• Welcome to TorchKGE’ s documentation! (for the labs)

Useful materials:

• PyKEEN is an alternative Python API. It is similar and may be more up-to-date than TorchKGE.

Session 11: Graph Neural Networks (GNNs)

Themes:

• Graph neural networks
  • recurrent/recursive, convolutional, GATs
• Question answering with GNNs (QA-GNN)
• Open KGs:
  • WordNet, BabelNet, ConceptNet

Mandatory readings:

• Section 5.3 Graph neural networks in Hogan et al.

Useful materials:

• The QA-GNN paper
• ConceptNet: An open, multilingual knowledge graph
• PyG Documentation: PyG (PyTorch Geometric) is a library built upon PyTorch to easily write and train Graph Neural Networks (GNNs) for a wide range of applications related to structured data.

Session 12: KGs and LLMs

Themes:

• Large Language Models (LLMs)
• Combining KGs and Large Language Models (LLMs)
  • retrieval augmented knowledge fusion
  • end-to-end KG construction
  • LLM-augmented KG to text generation
  • KG-LLM synergy

Mandatory readings:

• Chapter 8 Completion + Correction in Hogan et al.

Useful materials:

• Pan et al. (2024) Unifying large language models and knowledge graphs: A roadmap
• Vaswani et al. (2017) Attention is all you need
• Hitzler et al. (2022) Neuro-symbolic approaches in artificial intelligence

 

INFO216, UiB, 2017-2024, Andreas L. Opdahl (c)