Prof. Dr. Bernhard Steffen

Bernhard Steffen
Chairholder of Programming Systems at TU Dortmund University.

Department: Computer Science
Office: Room 102 (Otto-Hahn-Str. 14)
Address: TU Dortmund, 44221 Dortmund, Germany
Phone: +49 (231) 755-5800
Email: bernhard.steffen@tu-dortmund.de
ORCiD: 0000-0001-9619-1558

Google
Scholar
dblp
TU
Dortmund
Wikipedia

Research 🔗

My research is characterized by continuous tool development:

All tools have been published at international events and the tools highlighted in bold face are currently in use and available open source.

The interplay of the methods and tools captures various meta-levels:

  1. Language Workbench-level: The design of domain-specific languages for supporting low-code development as well as languages designed for verifiability, testability, learnability, and explainability. The point here is to guarantee properties via meta-level reasoning.
  2. IDE-level: Property-preserving transformation approaches to support correctness by construction, where the full semantics need to be preserved. This includes systematic verification benchmark generation, ensuring the satisfaction of pre/post conditions or LTL formulas, and aspect-specific view generation for diagnosis. The latter may even provide model explanations or outcome explanations for computational structures like Random Forests and Deep Neural Networks.
  3. Runtime: Behavior/observation-based inference of models and computational structures, comprising automata learning and other machine learning techniques such as Random Forests and Neural Networks. This inference may be active, explicitly querying the artifact to be analyzed, as typical for automata learning, or passive, relying on provided data sets, as typical in most Neural Network scenarios.

This interplay becomes particularly apparent in neuro-symbolic scenarios where code generated by LLMs is integrated into traditional software development and validated at runtime with techniques based on automata learning and model checking.

Inspired by the recent advances of LLMs, their enormous potential, and their societal impact, I founded the international event AISoLA (https://aisola.org/) to encourage interdisciplinary discussions among computer scientists, philosophers, psychologists, politicians, and lawyers. A corresponding state-of-the-art Lecture Notes in Computer Science volume (LNCS 15000) with interviews of experts in these fields will appear later this year. Because of the rapid development in the field, plan to organize AISoLA yearly in the next years.

Positions 🔗

Education 🔗

Teaching 🔗

I have taught Formal Methods in System Design (and related subjects) for 30 years as a professor. I introduced game-based learning for teaching strategy planning using Connect 4 and wrote an introductory book Foundations of Advanced Informatics and the German version Grundlagen der höheren Informatik.

Scientific Meetings, Journals & Community Activities 🔗

I have served on more than 150 Program Committees and over 30 times as chair, on numerous Steering Committees and in Editorial Boards.

I am (co-)founder of the

I am an editorial board member of Springer’s Innovations in Software and Systems Engineering: A NASA Journal, and have been a member of the editorial board of the ACM Transactions on Programming Languages and Systems (TOPLAS), Kluwer’s / Springer’s Formal Methods in System Design, and Springer’s Software-Concepts and Tools.

I am the longest serving editor of the LNCS (Lecture Notes in Computer Science), the leading proceedings series in computer science, for the sub-libraries Theoretical Computer Science, Programming Techniques and Software Engineering and Advanced Research in Computing and Software Science, and I am member of the GI, the ACM and EASST.

Fellowships and Awards 🔗

Supervision of Graduate Students and Postdoctoral Fellows 🔗

I (co-)supervise(d) over 40 PhD students of which 34 already completed their PhD studies successfully. 14 became associate/full professors:

Institutional Responsibilities 🔗