Understanding Modelling Assistance in Low-Code/No-Code Tools: An Empirical Study and the AssiStarMe Framework - Raw data and clustering script

This artifact contains the research protocol, raw data, literal quotes, scripts, demographic files, and clustering results related to the focus-group study reported in the paper "Understanding Modelling Assistance in Low-Code/No-Code Tools: An Empirical Study and the AssiStarMe Framework." Accepted at the main conference in MODELS2026. The artifact supports reuse, inspection, and replication of the empirical analysis presented in the paper. It includes the data and scripts used to generate the topic-based clusters and the final human-in-the-loop clusters reported for the paper’s research questions. The artifact contains the following files: Clusters.xlsx Contains the final clusters for RQ1 and RQ2 after human-in-the-loop refinement based on the topic-based clustering output. The file includes four sheets, each corresponding to one set of clusters, ranging from challenges to modelling-assistance features. Each sheet contains the cluster name/code, a cluster description, and the textual documents assigned to the cluster. Initial topic-based clustering results The artifact includes four CSV files containing the first version of the clusters generated with BERTopic before human-in-the-loop refinement: challenges_first_cluster.csv features_helped_first_cluster.csv features_hindered_first_cluster.csv missing_features_first_cluster.csv These files are provided to support reproducibility, since topic-clustering algorithms may produce slight variations in topic order, labels, or identifiers across executions. topic-based-clustering-script.ipynb Jupyter Notebook used to perform the BERTopic-based topic clustering over the raw focus-group data. The notebook includes the required Python dependencies and can be executed to regenerate the initial topic-based clustering outputs. Raw data files The raw data files contain the textual contributions extracted from the focus groups. Each file includes a session identifier indicating whether the contribution came from GI or GII. raw-data-challenges.xlsx: contributions related to challenges identified during the focus groups. raw-data-features-that-helped.xlsx: contributions describing modelling-assistance features that helped participants. raw-data-features-that-hindered.xlsx: contributions describing modelling-assistance features that hindered participants. raw-data-missing-features.xlsx: contributions describing modelling-assistance features that participants considered missing. Demographics The artifact includes Excel/CSV files containing anonymised participant demographic information for the focus-group sessions GI and GII. These files provide contextual information about the participants while preserving anonymity. Reproducibility notes To repeat the results, users can inspect the raw data files, execute topic-based-clustering-script.ipynb, compare the generated CSV files with the included initial topic-based clustering results, and then inspect Clusters.xlsx, which contains the final human-refined clusters used in the paper. Due to the stochastic nature of BERTopic and its underlying models, minor differences in topic ordering, labels, or identifiers may occur across executions. The final results reported in the paper should therefore be compared primarily against Clusters.xlsx. Ethical and legal considerations The artifact is intended for research transparency, review, reuse, and replication. The included focus-group data should be used only for research purposes. Users should not attempt to re-identify participants or infer personal information from the anonymised data.