New Publications from Our Group in Neurosymbolic AI
We are happy to announce that we have two new recently published papers from our Semantic Systems group in the field of neurosymbolic AI!
The first publication is a book chapter entitled “Analysing Objectives of Auxiliary Inputs in Semantic Web Machine Learning Systems” [link-to-paper], as part of the “Handbook on Neurosymbolic AI and Knowledge Graphs” published by IOS Press. This chapter was written in a collaboration with colleagues from Semantic Web Company (now merged into Graphwise.ai) and University of Mannheim. This chapter investigates the objectives behind incorporating auxiliary inputs, especially symbolic artefacts in neuro-symbolic systems. Through an analysis of almost 300 research papers, it explores how auxiliary symbolic inputs are used for various objectives –including enhancing performance, improving efficiency, and supporting interpretability–, and examines how these objectives relate to AI system architecture, tasks, and application domains.
The second publication is an article on Neurosymbolic Artificial Intelligence Journal entitled “Knowledge Engineering in the Age of Neurosymbolic Systems” [link-to-paper]. This article argues that the current rapid rise of AI technologies marks a new phase in knowledge engineering (KE), termed Neurosymbolic Knowledge Engineering. Building on the result of a systematic mapping study of systems, the authors present data-driven evidence of this emerging field, highlighting common system patterns, frequently used ML models, and Semantic Web resources for such neurosymbolic knowledge engineering. The article also outlines key challenges, such as development of new methodologies and addressing system maturity, to guide future research in this evolving area.
