IFIP TC12 - Artificial Intelligence

WG 12.1 - Knowledge Representation and Reasoning in Cognitive and Neuro-Symbolic AI (NSAI)

The working group 12.1 maintains the webpage KRportal which is a community platform for research on knowledge representation and reasoning.

Officers

Chair

This email address is being protected from spambots. You need JavaScript enabled to view it., BRGM and Lecturer at Université Évry Paris-Saclay France

Vice-Chair

Dr. Sanju Tiwari, Senior Researcher at TIB – Leibniz Information Centre for Science and Technology, Hannover, Germany, and Professor at Sharda University

Secretary

Dr. Amna Dridi, Senior Lecturer in Data Science, Birmingham City University

Background and Context

Knowledge Representation and Reasoning (KRR) is a foundational area of Artificial Intelligence concerned with the formal representation of knowledge and the computational mechanisms required to reason over it [1]. By explicitly representing entities, concepts, relations, constraints, and rules, KRR enables intelligent systems to derive knowledge beyond what is directly observed or stated.

The rapid development of machine learning and deep learning has substantially expanded the capabilities of AI systems to learn representations and patterns from data. In parallel, symbolic approaches provide explicit semantics, domain knowledge, constraints, and reasoning mechanisms. These paradigms address complementary aspects of intelligence: learning from data and reasoning with structured knowledge.

Knowledge Graphs (KGs) have become an important paradigm at the intersection of KRR, data management, and AI. Combined with ontologies, rules, and constraints, they support explicit knowledge representation and deductive reasoning, while recent approaches based on embeddings, graph neural networks, rule learning, and other inductive methods enable knowledge completion, prediction, and pattern discovery [2].

These developments have strengthened the interaction between symbolic and neural AI. Neuro-Symbolic AI (NSAI) investigates this interaction by combining machine learning with symbolic knowledge representation and reasoning [3]. Key research questions include how symbolic knowledge can guide learning, how neural methods can support knowledge acquisition and refinement, and how neural and symbolic reasoning mechanisms can be effectively integrated. These questions are closely related to relational reasoning, constraints, interpretability, and explainability [4].

This convergence is particularly relevant to Cognitive AI, where intelligent systems are expected to integrate perception, learning, knowledge representation, and reasoning. KRR can provide the explicit knowledge structures required to organize and exploit information acquired from data and perception, while learning mechanisms can support the acquisition, completion, and evolution of such knowledge.

Research at this intersection therefore raises challenges across several complementary levels: representation, including symbolic, neural, and hybrid representations; reasoning, including symbolic, neural, commonsense, and hybrid reasoning; the learning–reasoning interface, including knowledge acquisition, refinement, rule learning, and representation learning; and the system level, including cognitive agents and cognitive digital twins.

Within IFIP Technical Committee 12 on Artificial Intelligence, WG12.1 “Knowledge Representation and Reasoning in Cognitive and Neuro-Symbolic AI” provides a scientific forum for bringing together researchers from KRR, Semantic Web and Knowledge Graphs, machine learning, cognitive AI, and neuro-symbolic AI.

The Working Group focuses on the interaction between knowledge representation, learning, perception, and reasoning, covering symbolic, neural, and hybrid approaches. Knowledge Graphs constitute an important component of this scope, but the WG is not restricted to Knowledge Graph technologies. Its broader objective is to investigate how explicit and learned knowledge representations, together with reasoning mechanisms, can contribute to cognitive and neuro-symbolic AI.

Scientific Positioning

WG12.1 is positioned at the intersection of four complementary research dimensions:

• Knowledge Representation: ontologies, Knowledge Graphs, rules, axioms, and symbolic, neural, and hybrid representations;
• Reasoning: symbolic, Knowledge Graph, neural, commonsense, and hybrid reasoning mechanisms;
• Neuro-Symbolic AI: integration of neural learning with symbolic knowledge representation and reasoning;
• Cognitive AI: integration of perception, learning, knowledge, and reasoning within intelligent systems, including cognitive agents and cognitive digital twins.

The central scientific question of the WG is therefore:

How can knowledge representation and reasoning be integrated with learning and perception to support the development of cognitive and neuro-symbolic AI systems?

Aims

The aim of this WG is to extend the previous activity focused on Knowledge Representation and reasoning with modern AI techniques for better efficacity with less environmental impacts. The main objective of WG12.1 is to bring together researchers and practitioners working on Knowledge Representation and Reasoning, Knowledge Graphs, Cognitive AI, and Neuro-Symbolic AI, fostering international collaboration across symbolic, neural, and hybrid approaches.

The WG aims to advance the theoretical foundations, methods, and techniques for representing, acquiring, refining, and reasoning with knowledge, and to investigate their integration with learning and perception in cognitive and neuro-symbolic AI systems. Several activities may be included : scientific meetings, workshops, special tracks, panels, and other scientific events addressing the topics covered by the Working Group.

Scope

The scope of the Working Group’s activities includes (but is not restricted to) the following:   

• Knowledge representation and reasoning;

• Ontologies, knowledge graphs, rules, and axioms;

• Symbolic, neural, and hybrid knowledge representations;

• Symbolic, neural, and neuro-symbolic reasoning;

• Knowledge graph reasoning and representation learning;

• Rule learning and neuro-symbolic rules;

• Commonsense and relational reasoning;

• Knowledge acquisition, evolution, and refinement;

• Integration of perception, learning, knowledge, and reasoning;

• Interpretability and explainability;

• Cognitive ai architectures and systems;

• Cognitive and knowledge-driven agents;

• Knowledge-driven and explainable storytelling;

• Cognitive digital twins.

Future Plans

Workshop on “Knowledge Representation and Reasoning for Cognitive and Neuro-Symbolic AI” at IJCAI 2027

Workshop on “Semantic Foundations for Cognitive Digital Twins” at ESWC 2027

Edited Book on “Semantic Foundations and neuro-symbolique IA for Cognitive Digital Twins” with Springer 

Apply for:

• COST Action on “Semantic and Neuro-Symbolic Foundations for Cognitive Digital Twins” (SemCognDT) Funding Documents & Guidelines | COST deadline 26 october 2026
  
• Scientific Partnership with the International Knowledge Graph and Semantic Web Conference (KGSWC)
• MSCA Doctoral Network on “Multimodal Knowledge Graph Construction from Heterogeneous Information Sources”  deadline 24 novembre 2027

Members