Skip to main content
Nico Potyka

Dr Nico Potyka



School of Computer Science and Informatics


I am a Lecturer in the Knowledge Representation and Reasoning group at Cardiff University and affiliated with the Computational Logic and Argumentation group at Imperial College London as an Honorary Research Fellow. 

My research revolves around the following questions:

  • How can we represent and reason about knowledge?
  • How can we deal with uncertain and inconsistent knowledge?
  • How can we guarantee robustness and correctness of autonomous systems?
  • How we can explain the decisions of autonomous systems both faithfully and comprehensibly to a user?

I believe that the answer requires hybrid systems that combine the transparency and verifiability of symbolic methods with the flexibility and efficiency of machine-learning methods.

I am also interested in knowledge transfer and collaborations with industry partners.


Programming Libraries

For lack of a better place, I refer to some libraries here that I implemented together with colleagues over the years.

Attractor (Java): The Attractor library allows modeling and solving Quantitative/Gradual Argumentation problems. Arguments can be modeled in a directed graph, where nodes represent arguments and edges attack or support relationships between them. Every argument has an initial weight that represent an apriori strength when all other arguments are ignored. Reasoning algorithms assign a final strength to every argument based on its initial weight and the final strength of its attackers and supporters. Attractor supports various semantics and computes the final strength values by viewing the reasoning problem as a dynamical system that can be solved by numerical methods in polynomial-time (linear time for acyclic graphs).

ProBabble (Java): ProBabble allows modeling and reasoning about Probabilistic Argumentation problems. Similar to Gradual Argumentation Frameworks, argumentation problems are described by directed graphs. However, relationships between arguments are described by probabilistic constraints. ProBabble supports linear atomic constraints that allow representing many interesting relationships and reasoning in polynomial-time.

Uncertainpy (Python): Uncertainpy started as a Python version of Attractor but now contains additional tools that can be used for Quantitative Argumentation with weights or probabilities. As a rule of thumb, the Java implementations are faster but Uncertainpy is better suited for rapid prototyping.

Log4KR (Java): Log4KR is a Java library that allows modeling and reasoning about Probabilistic Reasoning problems. It supports propositional and relational probabilistic logic including probabilistic conditional logics. In order to deal with inconsistencies, it also provides implementations of inconsistency measures and paraconsistent reasoning algorithms. The library can be downloaded as part of the KReator project.





  • Xiong, B., Potyka, N., Tran, T., Nayyeri, M. and Staab, S. 2022. Faithful embeddings for EL++ knowledge bases. Presented at: International Semantic Web Conference (ISWC 2022), Hangzhou, China, 23-27 October 2022The Semantic Web – ISWC 2022, Vol. 13489. Lecture Notes in Computer Science Vol. 13489. Springer Cham pp. 22-38., (10.1007/978-3-031-19433-7_2)





Book sections

  • Hunter, A., Polberg, S., Potyka, N., Rienstra, T. and Thimm, M. 2021. Probabilistic argumentation: a survey. In: Gobbay, D. et al. eds. Handbook of Formal Argumentation, Volume 2., Vol. 2. College Publications, pp. 397-444.



Explainable AI

Automatic decision making is increasingly driven by black-box machine learning models. However, the opaqueness of these models raises questions about fairness, reliability and safety. For example, research in adversarial machine learning demonstrated that black-box models can be brittle and minor changes in the inputs can result in catastrophically different outputs that can be a severe risk in safety-critical applications like autonomous driving. Similarly, the current black-box nature of many models makes it impossible to guarantee that the model did not learn sexist, racial or other undesirable biases. Explainable AI aims at making autonomous systems more transparent. For example, by designing systems that are interpretable or by making the mechanics of black-box models more transparent.

Selected Publications

  • Nico Potyka, Yuqicheng Zhu, Yunjie He, Evgeny Kharlamov, Steffen Staab: Robust Knowledge Extraction from Large Language Models using Social Choice Theory. AAMAS 2024: To appear. Preprint
  • Francesco Leofante, Nico Potyka: Promoting Counterfactual Robustness through Diversity. AAAI 2024: To appear. Preprint
  • Nico Potyka, Xiang Yin, Francesca Toni: Explaining Random Forests using Bipolar Argumentation and Markov Networks. AAAI 2023: 9453-9460. Download
  • Hamed Ayoobi, Nico Potyka, Francesca Toni: SpArX: Sparse Argumentative Explanations for Neural Networks. ECAI 2023: 149-156. Download
  • Nico Potyka: Interpreting Neural Networks as Quantitative Argumentation Frameworks. AAAI 2021: 6463-6470. Download

Knowledge Graphs and Description Logics

Knowledge graphs can be seen as simple databases that represent data in the form of (subject, predicate, object) triples. Popular examples include DBpedia and YAGO that contain hundreds of millions of facts that can be used in intelligent systems. Decription logics are formalisms that allow reasoning about the data in the knowledge graph in order to infer new information that is not explicitly stored. Knowledge graph embeddings aim at representing knowledge graphs as vectors (similar to how word embeddings represent words) and can be used as a standalone tools for plausible reasoning or to inject background knowledge into machine learning models. Ontology and rule embeddings refine standard knowledge graph embeddings by taking logical relationships into account that can improve the overall embedding.

Selected Publications

  • Bo Xiong, Nico Potyka, Trung-Kien Tran, Mojtaba Nayyeri, Steffen Staab:
    Faithful Embeddings for Eℒ++ Knowledge Bases. ISWC 2022: 22-38. Download
    (Best Student Paper Award)
  • Bo Xiong, Shichao Zhu, Nico Potyka, Shirui Pan, Chuan Zhou, Steffen Staab:
    Pseudo-Riemannian Graph Convolutional Networks. NeurIPS 2022. Download
  • Rafael Peñaloza, Nico Potyka: Towards Statistical Reasoning in Description Logics over Finite Domains. SUM 2017: 280-294. Download

Computational Argumentation

Computational argumentation studies methods to represent and reason about arguments that naturally occur in online discussions, political debates or general decision problems. As opposed to the assumptions of classical logic, argumentation problems are naturally filled with contradicting arguments, so that arguments often cannot be declared as definitely true or definitely false, but rather as acceptable or non-acceptable. Argumentation formalisms can be roughly divided into structured approaches that take the logical structure of arguments into account and abstract approaches that abstract from the content of arguments and focus on their relationships. Furthermore, we can distinguish qualitative approaches that focus on identifying acceptable sets of arguments and quantitative approaches that quantify the acceptability of arguments.

Selected Publications

  • Anthony Hunter, Nico Potyka: Syntactic Reasoning with Conditional Probabilities in Deductive Argumentation. Artificial Intelligence 321. 2023. Download
  • Nico Potyka: Abstract Argumentation with Markov Networks. ECAI 2020: 865-872. Download
  • Anthony Hunter, Sylwia Polberg, Nico Potyka: Updating Belief in Arguments in Epistemic Graphs. KR 2018: 138-147. Download
  • Nico Potyka: Continuous Dynamical Systems for Weighted Bipolar Argumentation. KR 2018: 148-157. Download

Probabilistic Reasoning

Probabilistic reasoning approaches go beyond classical reasoning by evaluating claims using probabilities rather than the classical truth values True and False. One of the most popular probabilistic reasoning approaches are Probabilistic graphical models, which represent random variables and their relationships in a graphical structure and exploit independencies in order to perform learning and inference more efficiently. Another interesting approach are Probabilistic logics that combine classical logic and probability theory to allow for automated reasoning with uncertain information that naturally occurs in applications like medical diagnosis or legal reasoning. The combination of the two allows describing reasoning problems more naturally than in pure probability theory (using logical formulas rather than abstract events or random variables) and more accurately than in pure logic (replacing the truth values 0 and 1 by the probability interval from 0 to 1). The area of Statistical relational artificial intelligence brings together ideas from probabilistic graphical models, probabilistic logics and logic programming.

Selected Publications

  • Nico Potyka: A polynomial-time fragment of epistemic probabilistic argumentation. International Journal of Approximate Reasoning 115: 265-289 (2019). Download
  • Nico Potyka, Matthias Thimm: Inconsistency-tolerant reasoning over linear probabilistic knowledge bases. International Journal of Approximate Reasoning 88: 209-236 (2017). Download
  • Nico Potyka, Erman Acar, Matthias Thimm, Heiner Stuckenschmidt: Group Decision Making via Probabilistic Belief Merging. IJCAI 2016: 3623-3629. Download

Inconsistency Tolerance

Both classical and probabilistic logic are brittle in the sense that contradictory information can render logical consequences meaningless. There are various ways to overcome this problem. Repair-operators try to repair an inconsistent knowledge base while maintaining as much of the consistent information as possible. Another approach is to design inconsistency-tolerant reasoning approaches that can derive non-trivial results even if the knowledge base is inconsistent. Inconsistency measures allow quantifying the degree of inconsistency in order to make a more informed choice about the right tool.

Selected Publications

  • Nico Potyka, Matthias Thimm: Probabilistic Reasoning with Inconsistent Beliefs Using Inconsistency Measures. IJCAI 2015: 3156-3163. Download
  • Nico Potyka: Reasoning over Linear Probabilistic Knowledge Bases with Priorities. SUM 2015: 121-136. Download
  • Nico Potyka: Linear Programs for Measuring Inconsistency in Probabilistic Logics. KR 2014. Download





Cardiff University (since 2023)

I am the Undergraduate Programme Lead for Computer Science and am teaching the following modules

  • Computational Thinking: together with Dr Daniela Tsaneva (Autumn Term 2023).


University of Stuttgart (2020 - 2022)

I taught several modules related to Artificial Intelligence including

  • Knowledge Graphs: together with Professor Steffen Staab (Spring Term 2021),
  • Introduction to Artificial Intelligence (Autumn Term 2020),
  • Commonsense Reasoning (Seminar): together with Teresa Kraemer (Autumn Term 2020),
  • Knowledge Graphs (Seminar): together with Teresa Kraemer and Professor Steffen Staab (SpringTerm 2020).


University of Osnabrueck (2016 - 2020)

I taught several modules related to Artificial Intelligence including

  • Introduction to Artificial Intelligence and Logic Programming: together with Dr Tobias Thelen (Spring Term 2017 - 2019),
  • Methods of Artificial Intelligence: together with Professor Kai-Uwe Kuehnberger (Autumn Term 2016 - 2019),
  • Basic Methods of Probabilistic Reasoning (Spring Term 2016 - 2018, Autumn Term 2019),
  • Rational Reasoning in Multi-agent Systems (Autumn Term 2016 - 2018),
  • Selected Topics in Nature-inspired Algorithms (Autumn Term 2017, Spring Term 2018, 2019),
  • Time Series Analysis and Forecasting (Autumn Term 2019),
  • Selected Topics in Constraint Programming (Seminar, Spring Term 2017),
  • Conceptual Spaces - Applications and Learning (Seminar): together with Dr Lucas Bechberger and Dr Eleni Gregoromichelaki (Spring Term 2017).


University of Hagen (2011 - 2015)

I was teaching assistant for the annual modules

  • Methods of Knowledge Representation and Reasoning,
  • Logical and Functional Programming,
  • Knowledge-based Systems,
  • Deductive Systems,
  • Knowledge Representation and Reasoning (Seminar).



Before joining Cardiff University as a Lecturer in 2023, I had previous academic positions at

and spent research visits at

Before studying Computer Science (2005-2010), I worked in Electronics (2000-2001) and Logistics (2001-2004). I also worked as a Software Developer on Enterprise Resource Planning (2010-2011) between my Master and my PhD.

Scientific Activities

Honours and awards

Speaking engagements

  • Cardiff NLP Summer Workshop, 2023: Panel Discussion on NLP Research in the Age of LLMs (Panel Member)
  • University of Bergen, 2023: Concept Embeddings (Invited Talk)
  • St Paul's Girls School, London, 2023: Complexity of Algorithms (Tutorial)
  • University of Hagen, 2022: Markov Networks for Learning and Reasoning about Argumentation Problems (Invited Talk)
  • The 8th Workshop on Probabilistic Logic Programming (PLP 2021): From Probabilistic Programming to Probabilistic Argumentation (Invited Talk)
  • German Conference on Artificial Intelligence (KI 2020): Explainable and Computationally Efficient Decision Making with Quantitative Abstract Argumentation Frameworks (Tutorial)
  • German Conference on Artificial Intelligence (KI 2019): Modeling and Solving Weighted Bipolar Argumentation Problems (Tutorial)
  • Technical University of Dresden, 2019: Probabilistic Reasoning with Conflicting Information (Invited Talk)
  • Workshop on Ontologies, Uncertainty, and Inconsistency Handling, 2018: Knowledge Representation and Reasoning with Nilsson-style Probabilistic Logics (Tutorial)
  • German Conference on Artificial Intelligence (KI 2017): Knowledge Representation and Reasoning with Nilsson-style Probabilistic Logics (Tutorial)
  • University of Mannheim, 2015: Probabilistic Reasoning with Consistent and Inconsistent Information (Invited Talk)

Committees and reviewing

I serve on the Program Committees of major and smaller specialized conferences and workshops including the conferences

  • IJCAI: International Joint Conference on Artificial Intelligence (since 2017),
  • AAAI: AAAI Conference on Artificial Intelligence (since 2020),
  • KR: International Conference on Principles of Knowledge Representation and Reasoning (since 2022),
  • ICML: International Conference on Machine Learning (since 2022),
  • NeurIPS: Conference on Neural Information Processing Systems (since 2022),
  • ECSQARU: European Conf. on Symbolic and Quantitative Approaches to Reasoning with Uncertainty (since 2019),
  • SUM: International Conference on Scalable Uncertainty Management (since 2020),
  • COMMA: International Conference on Computational Models of Argument (since 2020),
  • WSDM: International Conference on Web Search and Data Mining (2021-2023),
  • KDD: ACM SIGKDD Conference on Knowledge Discovery and Data Mining (2021-2023).

I also serve as a reviewer for major and smaller specialized journals including

  • AIJ: Artificial Intelligence Journal,
  • JAIR: Journal of Artificial Intelligence Research,
  • IJAR: International Journal of Approximate Reasoning.

Research themes


  • Artificial intelligence
  • Computational logic and formal languages
  • Knowledge representation and reasoning
  • Explainable AI
  • Interpretable Machine Learning