Trends and Perspectives

Knowledge Graphs and Neurosymbolic AI

Knowledge Graphs provide a robust framework for organizing and representing large-scale knowledge in machine-readable formats. Their flexibility and scalability in capturing knowledge makes them invaluable across diverse domains, fostering seamless integration with Machine Learning techniques. This integration enhances their relevance for traditional AI applications and positions them as crucial components in the emerging field of Neurosymbolic AI. At its core, Neurosymbolic AI bridges the strengths of symbolic reasoning and neural network learning.

In the current ever-evolving landscape of AI, this workshop aims to unravel the potential lying at the intersection of Knowledge Graphs and Neurosymbolic AI. Neurosymbolic AI/Machine Learning techniques can be applied to construct and refine Knowledge Graphs, contributing to their ongoing evolution. Simultaneously, leveraging Knowledge Graphs for learning in Neurosymbolic AI allows intelligent systems to draw insights from structured and unstructured symbolic representations, paving the way for more informed AI models and improving interpretability and transparency in their decision-making process.

We invite papers that explore these synergies as well as any other combinations, seeking to gain a better understanding of how Knowledge Graphs and Neurosymbolic AI influence and benefit each other. Application papers and extended abstracts of published papers are also welcome.

Topics of interests

Topics of interest include, but are not limited to :

  • Neurosymbolic AI for Knowledge Engineering.
  • Machine Learning techniques for creating, improving or aligning Knowledge Graphs.
  • Knowledge Infusion in Machine Learning algorithms.
  • Knowledge Graphs quality and its influence on Neurosymbolic AI systems.
  • Knowledge Graphs for trustworthy Neurosymbolic AI systems.
  • Knowledge Graphs for explainable AI.
  • Knowledge Representation and Reasoning using Deep Neural Networks.
  • Methods, systems, and techniques using Symbolic AI for the development of Explainable AI.
  • Utilizing ontologies for enhancing Neurosymbolic AI on the Web.
  • Applications of Neurosymbolic AI and Knowledge Graphs in Industry.
  • Applications of Neurosymbolic AI in domains such as medicine, biology, IoT, search, security and others.

Keynote Speakers

We are happy to announce that Alexander Schindler from Austrian Institute of Technology (AIT) and Axel Polleres from Wirtschaftuniversität Wien (WU Wien) are our keynote speakers for the workshop!

Keynote 1 (Tuesday): Alexander Schindler

Title: A hybrid approach to Disinformation Detection

Identifying disinformation is a difficult task because it can occur in different forms, modalities, languages, and contexts. Disinformation also often arises from the contradiction between written and visual content. It may distort or deny facts, or it may aim to emotionalize. To recognize disinformation, it is necessary to consider current and culturally specific contexts. In addition, we are constantly operating on the edge of fundamental rights such as freedom of expression and freedom of the press. Years of continuous research on this topic have shown that an end-to-end solution with a single model is illusory. In this talk, I will present a hybrid artificial intelligence approach that uses a combination of expert machine learning models to enrich task-specific metadata stored in a graph-based knowledge representation as a basis for performing high-level analysis, interpretation, or prediction, and combinations of language models and taxonomies to partition data streams into relevant categories to facilitate situational awareness.

Keynote 2 (Wednesday): Axel Polleres

Title: The different “Shapes” of RDF(S) and OWL: A Fragmented History

Since the introduction of the Semantic Web in the late 90s, schema and ontology languages to describe the schema of what we now call “Knowledge Graphs” have played a central role. The semantic basis of these ontology languages have - historically - been based on formalisms such as Frame Logic, Description Logics as well as Datalog. The syntactic representations of Schema axioms are integrated in Knowledge Graphs by representations of axioms in RDF, using the W3C standardized RDF, RDFS and OWL vocabularies. OWL and RDFS can therefore be both seen as logical languages, but also simply as RDF vocabularies, a constrained use of which allows us to “encode” terminological axioms as part of an RDF (knowledge) graph. Yet, an unconstrained use of these vocabularies yields obviously “unintuitive” graphs.

In this keynote talk we would like to discuss two questions, namely: (a) is there too much syntactic freedom in RDF and OWL? (b) (how) can useful syntactic fragments of OWL and RDFS usage be captured by constraints and shapes? In the course of (b) we also aim at providing an “historical” overview of (semantic and syntactic) OWL and RDFS fragments from the literature.

Workshop Program

Tuesday

[13:30-15:00] Session I

  • Workshop Opening (Chairs)
  • [KEYNOTE] Alexander Schindler: “A hybrid approach to Disinformation Detection”.
  • [Full Paper] Majlinda Llugiqi, Fajar J. Ekaputra, Marta Sabou: “Leveraging Knowledge Graphs for Enhancing Machine Learning-based Heart Disease Prediction” [Full Paper] [Slide].

[15:30-17:00] Session II

  • [Full Paper] Tolga Çöplü, Arto Bendiken, Andrii Skomorokhov, Eduard Bateiko, Stephen Cobb, Joshua J. Bouw: “Prompt-Time Symbolic Knowledge Capture with Large Language Models” [Full Paper].
  • [Extended Abstract] Marius-Constantin Dinu, Claudiu Leoveanu-Condrei, Markus Holzleitner, Werner Zellinger, Sepp Hochreiter: “SymbolicAI: A framework for logic-based approaches combining generative models and solvers” [Extended Abstract].
  • [Short Paper] Michele Collevati, Thomas Eiter, Nelson Higuera: “Leveraging Neurosymbolic AI for Slice Discovery” [Short Paper].
  • [Extended Abstract] Artem Revenko: “Supplementary Objectives: Analysing The Motivations Behind Semantic Web Machine Learning System Design” [Extended Abstract].

Wednesday

[09:10-10:30] Session III

  • [KEYNOTE] Axel Polleres: “The different “Shapes” of RDF(S) and OWL: A Fragmented History”.
  • [Short Paper] Umutcan Serles, Ioan Toma: “LLMs Meet Knowledge Graphs For A More Transparent Conversational AI: Wien Energie Chatbot” [Short Paper].
  • [Extended Abstract] Albin Ahmeti, Robert David, Artem Revenko, Jan-Kees Schakel: “A Species Conservation Recommender System based on Knowledge Graph for Brown Bear Movement Prediction” [Extended Abstract].

[11:00-12:30] Session IV

  • [Short Paper] Alaa El-Ebshihy, Filip Kovacevic, Florina Piroi, Andreas Rauber: “Extending Content-based Scientific Knowledge Graphs with Research Results” [Short Paper].
  • [Extended Abstract] Aleksandar Pavlović, Emanuel Sallinger: “Expressive and Geometrically Interpretable Knowledge Graph Embedding” [Extended Abstract] [Slide].
  • [Short Paper] Simon Ott, Daria Liakhovets, Mina Schütz Medina Andresel, Mihai Bartha, Sven Schlarb, Alexander Schindler: “Knowledge Graphs and their Applications in Civil Security” [Short Paper].
  • Workshop Discussion and Conclusion (Chairs)

Important Dates

  • 2024-01-31: Paper submission deadline
  • 2024-02-28: Notification of acceptance
  • 2024-03-26: AIRoV Symposium

Submissions

Submissions can fall in one of the following categories:

  • Full research papers (8 pages)
  • Short research papers (4 pages)
    • Work-in-progress
    • Position papers
  • Extended abstract (2 pages)
    • Extended abstract of accepted conference/journal papers.
    • Project description / Lightning talks

Submit your contribution to this workshop on the AIROV’s CMT page, identified by the workshop acronym (KG-NeSy). The review process will be conducted in a single-blind review process. Further information on the format can be found here.

At least one of the authors of the accepted papers must register for the workshop (pre-conference only option) to be included into the workshop proceedings.

Selected papers from the workshop will be invited to the Special issue on Knowledge Graphs and Neurosymbolic AI of the NAI Journal.

Workshop Organizers

Program Committee