My research centers on Knowledge-based Diagnosis, Constraint Acquisition, Configuration Systems, Software Product Lines, and Explanations in AI, with recent work extending these foundations to Large Language Models for Software Product Line Engineering. The goal is to make knowledge-based systems — such as configuration systems and recommender systems — faster, more transparent, and more useful for end users, by combining classical constraint reasoning with modern learning-based techniques.
I am a team member of the following projects, led by Professor Alexander Felfernig (PI):
MIRROR — Material Improvement through Reflective Review of Outputs and Resources. A teaching innovation project funded by Projektfonds Lehre 2026 (TU Graz Vice-Rectorate for Teaching) that uses Large Language Models to analyze anonymized student submissions from the Introduction to Structured Programming course and propose targeted didactic improvements for the upcoming summer semester.
GenRE — Generative AI for Requirements Engineering. An FFG Bridge project (2024–2027) developing LLM-based techniques for requirements elicitation, quality assurance, and validation in software engineering. Industry partner: Morgendigital; external evaluation partners: Innovation Service Network and Uniquare.
OpenSpace — AI Techniques for Testing Highly-Variant Software. An FFG Bridge project (2021–2024) developing machine learning approaches for testing and debugging variability-intensive software, including automated analysis of variability models and the identification of faulty components and suboptimal parameterizations. Industry partner: Uniquare GmbH.
ParXCel — Machine Learning and Parallelization for Scalable Constraint Solving. An FFG Bridge project (2020–2023) integrating machine learning into constraint-based reasoning to enable personalized configuration, and parallelizing analysis operations such as conflict detection and diagnosis to boost performance. Industry partner: Combeenation GmbH.
I have developed (and contributed to) the following software and tools:
flamapy (Python) — open-source ecosystem for the automated analysis of feature models. Contributed FastDiagP, DirectDebug, and WipeOutR as plugins. FastDiagP is also accessible interactively through the browser-based environment flamapy.ide.
FMTesting (Java, FeatureIDE plug-in) — Eclipse plug-in for feature model testing and debugging, built on top of hiconfit-core. Integrates DirectDebug, WipeOutR, and AggregatedTest.
Restful Configurator Webservice (Java, REST API, Spring Boot) — Proprietary REST API for developing product configurators, developed for Combeenation GmbH within the ParXCel project. Built on top of hiconfit-core. Provides domain reduction, matrix factorization-based configuration and recommendation, option reordering via Value Variable Heuristics, and conflict and diagnosis detection.
DirectDebug (Java) — software package for the automated testing and debugging of variability models. Published in Software Impacts (2021).
FM2ExConf (Java) — converts feature models into executable Excel-based configurators, built on top of hiconfit-core. Supports anomaly detection and configuration explanation, making configuration accessible to non-IT stakeholders.
HiConfiT (Java) — High Performance Knowledge Based Configuration Techniques. A suite of open-source libraries (hiconfit-core) and command-line apps (KBStatistics, FMGen) for Knowledge-Based Configuration Systems. Used by FMTesting and FM2ExConf.