Paper accepted at ICSME 2026
26.06.2026
We are happy to share that our Registered Report "Training Dynamics of Neural Software Defect Predictors under Coupled Data-Quality Issues" by Emmanuel Charleson Dapaah, Philip Makedonski and Jens Grabowski has been accepted at the
42nd International Conference on Software Maintenance and Evolution (ICSME 2026). The conference will take place from 14 to 18 September 2026 in Benevento, Italy.
This study investigates how training-dynamics patterns from class imbalance, overlap, and their coupling can be characterized under interaction-aware conditions in deep learning-based Software Defect Prediction.
Article published at Innovations in Systems and Software Engineering
26.06.2026
We are happy to announce that our article
"From diagnosis to repair: A model-driven framework for root cause analysis of machine learning pipelines" is published online in the Journal of Innovations in Systems and Software Engineering. In the article, we propose a model-driven framework for root cause analysis and hyperparameter intervention that operates solely on structured descriptors. The framework uses three dataset-complexity meta-features, namely class overlap, class imbalance, and sparsity, together with learner hyperparameters. We evaluate the approach on two model families, Decision Trees (DT) and Multilayer Perceptrons (MLP). For each family, we construct a meta-dataset comprising 81,000 pipeline runs generated from 270 datasets and 300 hyperparameter configurations.
Funding for AI-generated code quality project
26.06.2026
We are pleased to announce that our proposal “A Reproducible Benchmark for AI-Generated vs. Human-Written Code: Comparing Quality and Test Demand” has been approved for funding through the Internal Call for Project Ideas 2026 at the Institute of Computer Science. The project will run in 2026 and aims to compare AI-generated and human-written code with respect to code quality and testing needs.
Paper accepted at EASE 2026
01.04.2026
We are happy to share that our paper Quality-Driven Selective Mutation for Deep Learning by Zaheed Ahmed, Emmanuel Charleson Dapaah, Philip Makedonski and Jens Grabowski has been accepted at the
30th International Conference on Evaluation and Assessment in Software Engineering (EASE 2026). The conference will take place from 9 to 12 June 2026 in Glasgow, United Kingdom.
This study presents a probabilistic framework for quantifying mutant quality along two complementary dimensions: resistance and realism. The framework enables the ranking and filtering of low-quality mutation-operator configurations, substantially reducing mutation cost while preserving the practical value of mutants for testing and debugging.
Paper accepted at JAWs 2026
01.04.2026
Paper accepted at HCII 2026
19.12.2025
We are pleased to announce that a paper from our research group has been accepted at the
28th International Conference on Human-Computer Interaction (HCII 2026). This year’s conference will take place from July 26th to 31st in Montreal, Canada. Our accepted paper is:
- From Promise to Proof: Assessing Replicability and Reproducibility in Interaction-Based Emotion Recognition Research by Carina Bieber, Patrick Harms, and Jens Grabowski
Paper accepted at SAM 2025
02.10.2025
We are happy to announce that our paper Model-Driven Root Cause Analysis for Trustworthy AI: A Data-and-Model-Centric Explanation Framework (by Emmanuel Charleson Dapaah and Jens Grabowski) has been accepted at the The 17th System Analysis and Modelling conference (SAM 2025). This paper presents a model-driven Root Cause Analysis framework that explains ML pipeline performance by attributing outcomes to interpretable factors spanning both data complexity and model configuration. Paper accepted at SAM 2025.
Latest TDL specifications published
14.08.2025
ETSI’s Methods for Testing and Specification (MTS) committee has published an update of its TDL specifications. These specifications are:
The outcomes of our continued work on the Test Description Language (TDL) within TTF T034 of the European Telecommunications Standards Institute (ETSI) have been published as updated versions of the specifications:
You can find the complete series of TDL specifications here.
Paper accepted at LOD 2025
25.04.2025
We are happy to announce that our paper Empirical Evidence for Data-Centric AI: A Comparative Study of Data Complexity and Hyperparameter Effects (by Emmanuel Charleson Dapaah and Jens Grabowski) has been accepted at the The 11th International Conference on Machine Learning, Optimization, and Data Science (LOD 2025). This paper presents a comprehensive empirical study comparing the relative influence of dataset complexity and hyperparameter settings on the performance of five widely-used classification algorithms: Random Forest, Support Vector Machine, Decision Tree, Adaptive Boosting, and Multi-Layer Perceptron. The findings reveal that data-centric factors—especially class overlap (N1)—consistently exert a far stronger impact on both bias and variance than hyperparameter settings.
New TTF for TDL Maintenance
03.03.2025
