Institute of Software Engineering and Artificial Intelligence

The Institute of Software Engineering and Artificial Intelligence is concerned with the theoretical, practical and applied aspects of software engineering. Research lies in the field of intelligent systems, formal verification and systematic testing of software, artificial intelligence, requirements engineering, recommender systems, optimization of industrial problems, agile software development processes, intelligent autonomous robots, innovative software engineering education, integrating AI techniques within coding education and DevOps processes, AI applications in understanding the cognition of octopuses contrasted with artificial neural networks, and investigates strategies to responsibly address the challenges posed by future AI superintelligence..

From Feature Models to Large Language Models: A Variability Engineering Perspective

José A. Galindo
(University of Seville, Spain)

April 23^rd, 2026  |  16:15
Location: Lecture Hall i11, TU Graz (Inffeldgasse 16b, 8010 Graz)

Abstract

Software Product Line Engineering has long relied on feature models as a formal backbone for managing variability,  capturing what configurations are possible, valid, and meaningful. But as systems grow in complexity and AI-assisted development becomes mainstream, new questions arise: can we automate the creation of realistic feature model corpora? Can we rely in stochastic AI techniques to reason over structured variability artifacts?

We will present the journey of DiversoLab (Universidad de Sevilla) through these questions. We will introduce flamapy, an open-source Python framework for automated analysis of feature models, and UVL (Universal Variability Language), the community-driven exchange format that now connects a broad ecosystem of tools. I then describe our work on using LLMs to generate syntactically valid and semantically coherent feature model instances from UVL prompts, revealing both the promise and the limits of code-oriented language models for domain-specific language generation.