Dataset: Understanding Usefulness in Developer Explanations on Stack Overflow

Description This replication package accompanies the study Understanding Usefulness in Developer Explanations on Stack Overflow. It provides the dataset and scripts needed to reproduce the analyses reported in the paper and to support further replication or extension. The dataset comprises 3,323 questions and 59,398 answers collected from Stack Overflow, enriched with metadata and computed features such as sentiment polarity, content structure metrics (e.g., length, code blocks, links), temporal information, and author-related attributes. It is used to investigate how structural, contextual, and social factors relate to the perceived usefulness of developer explanations. The package includes: Dataset: A processed JSON dataset with explanations, including sentiment (and politeness) annotations. Sentiment analysis: A paper-aligned BERT-only inference pipeline (paper_pipeline/) used for the sentiment polarity annotations, plus optional exploratory benchmarking material (exploratory_benchmarks/) provided for transparency and not used in the reported analyses. Statistical analysis: Correlation-based analysis scripts (Spearman, point-biserial, eta) and result exports to reproduce the tables and findings reported in the paper. A detailed README is included with folder structure and reproduction instructions. AuthorsMartin Obaidi, Kushtrim Qengaj, Hannah Deters, Jakob Droste, Marc Herrmann, Kurt Schneider, Jil Klünder CitationIf you use this dataset or the accompanying scripts, please cite:Obaidi, M., Qengaj, K., Deters, H., Droste, J., Herrmann, M., Schneider, K., Klünder, J.: Understanding Usefulness in Developer Explanations on Stack Overflow. In: Requirements Engineering: Foundation for Software Quality: 32nd International Working Conference (REFSQ 2026). ContactMartin Obaidi ([e-mail]) LicenseUnless otherwise noted, all data and scripts are licensed under the Creative Commons Attribution 4.0 International License (CC BY 4.0).