Dataset: How to Elicit Explainability Requirements? A Comparison of Interviews, Focus Groups, and Surveys

Dataset: How to Elicit Explainability Requirements? A Comparison of Interviews, Focus Groups, and Surveys Authors:Martin Obaidi, Jakob Droste, Hannah Deters, Marc Herrmann, Raymond Ochsner, Jil Klünder, Kurt Schneider Description This dataset accompanies the publication:How to Elicit Explainability Requirements? A Comparison of Interviews, Focus Groups, and Surveys(2025 IEEE 33rd International Requirements Engineering Conference, RE 2025) The dataset provides all materials, question sets, coding guidelines, and coded results from a study investigating how to elicit explainability requirements in real-world software development. Three data collection methods were compared: interviews, focus groups, and surveys. All data were collected in an organizational context. This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Grant No.: 470146331, project softXplain (2022–2025). Contents The dataset includes the following files: All_explanation_needs_coded.xlsxContains all coded explanation needs identified in interviews, focus groups, and surveys.Each need is categorized using an established taxonomy (see below).The file includes information on method, session, participant, taxonomy version (direct, delayed, without), and whether the need is unique/distinct. Main columns: Session/ID: Identifies source (interview, focus group, survey). Taxonomy usage: Version of taxonomy used (direct, delayed, without). Explanation type category / subcategory: Main and subcategory for each need, based on an extended taxonomy. Additional taxonomy columns: Fine-grained subcategories (e.g., time aspect, software feature, system aspect) used for precise grouping and counting of needs. Combined string: Concatenation of taxonomy fields, used to detect identical needs. Occurrence: Number of times this need appeared in the dataset. Taxonomy coding guidelines.pdfThe full coding guide, describing all categories and subcategories used for coding explanation needs, adapted from Droste et al. and extended by Obaidi et al.Each category includes a definition and practical example. Droste et al.: Droste, J., Deters, H., Obaidi, M. et al. Framing what can be explained - an operational taxonomy for explainability needs. Requirements Eng (2025).https://doi.org/10.1007/s00766-025-00440-xObaidi et al.: M. Obaidi, N. Voß, J. Droste, H. Deters, M. Herrmann, J. Fischbach, and K. Schneider, “Automating explanation need management in app reviews: A case study from the navigation app industry,” in ICSE-SEIP’25, 2025https://arxiv.org/abs/2501.08087 taxonomy.pdfA graphical overview of the taxonomy as provided to study participants. questions_fokus_group.xlsxAll questions used in the focus groups, as well as the measured time spent on each question and session.The file distinguishes between focus groups with direct, delayed, or without taxonomy introduction. questions_interviews.xlsxAll interview questions and measured times per question/session.Includes session details for each taxonomy version (direct, delayed, without). questions_survey.xlsxThe survey instrument, with all questions and measured times, organized by taxonomy version (without and delayed).Questions involving the introduction and application of the taxonomy are highlighted in blue for clarity. Anonymization and Privacy To comply with privacy and company requirements, all data have been fully anonymized: Company and software names have been removed or replaced by placeholders. Demographic and potentially identifying information has been deleted. Only non-sensitive, anonymized qualitative data is included. Usage and Citation This dataset can be used for: Secondary analysis of explanation needs in software engineering Methodological comparison of requirements elicitation techniques Development or validation of explainability taxonomies Training and education in qualitative coding and requirements engineering If you use this dataset, please cite the following publication: Obaidi, M., Droste, J., Deters, H., Herrmann, M., Ochsner, R., Klünder, J., Schneider, K. (2025).How to Elicit Explainability Requirements? A Comparison of Interviews, Focus Groups, and Surveys.2025 IEEE 33rd International Requirements Engineering Conference (RE). License This dataset is provided under the Creative Commons Attribution 4.0 International License (CC BY 4.0). Contact For questions regarding the dataset, please contact the corresponding author as listed in the publication.数据集：如何获取可解释性需求（How to Elicit Explainability Requirements）？——访谈、焦点小组与问卷调查的对比研究 作者：Martin Obaidi, Jakob Droste, Hannah Deters, Marc Herrmann, Raymond Ochsner, Jil Klünder, Kurt Schneider ## 数据集说明 本数据集配套发表于《如何获取可解释性需求？——访谈、焦点小组与问卷调查的对比研究》（2025年第33届IEEE国际需求工程会议（RE 2025））。 本数据集包含一项针对真实软件开发场景下可解释性需求（Explainability Requirements）获取方法的研究的全部相关材料、问题集、编码指南与编码结果。该研究对比了三种数据收集方法：访谈、焦点小组与问卷调查，所有数据均采集自企业组织场景。 本研究由德国研究基金会（Deutsche Forschungsgemeinschaft, DFG）资助，项目编号470146331，项目名称为softXplain（2022–2025）。 ## 数据集内容 本数据集包含以下文件： 1. **All_explanation_needs_coded.xlsx**：包含访谈、焦点小组与问卷调查中识别出的全部已编码解释需求。每项需求均基于既定分类体系（详见下文）完成分类。该文件包含数据来源方法、会话编号、参与者信息、分类体系版本（直接版、延迟版、无引导版）以及需求是否唯一/独特的相关字段。 主要列说明： - 会话/标识符（Session/ID）：标识数据来源类型（访谈、焦点小组或问卷调查）。 - 分类体系使用情况（Taxonomy usage）：所采用的分类体系版本（直接版、延迟版、无引导版）。 - 解释类型类别/子类别（Explanation type category / subcategory）：基于扩展分类体系的每项需求的主类别与子类别。 - 额外分类体系列（Additional taxonomy columns）：用于精准分组与统计需求的细粒度子类别（例如时间维度、软件功能、系统层面等）。 - 组合字符串（Combined string）：分类体系字段的拼接结果，用于识别重复需求。 - 出现频次（Occurrence）：该需求在数据集中出现的总次数。 2. **Taxonomy coding guidelines.pdf**：完整的编码指南，详述了用于编码解释需求的所有类别与子类别。该分类体系改编自Droste等人的研究，并由Obaidi等人扩展完善，每个类别均包含明确定义与实际应用示例。 参考文献： - Droste等人：Droste, J., Deters, H., Obaidi, M. et al. Framing what can be explained - an operational taxonomy for explainability needs. Requirements Eng (2025). https://doi.org/10.1007/s00766-025-00440-x - Obaidi等人：M. Obaidi, N. Voß, J. Droste, H. Deters, M. Herrmann, J. Fischbach, and K. Schneider, "Automating explanation need management in app reviews: A case study from the navigation app industry," in ICSE-SEIP’25, 2025 https://arxiv.org/abs/2501.08087 3. **taxonomy.pdf**：提供给研究参与者的分类体系图形化概览文档。 4. **questions_fokus_group.xlsx**：焦点小组使用的全部问题，以及每个问题与会话的耗时统计。该文件区分了采用直接版、延迟版或无引导版分类体系介绍的焦点小组会话。 5. **questions_interviews.xlsx**：所有访谈问题及每个问题/会话的耗时统计，包含各分类体系版本（直接版、延迟版、无引导版）对应的会话详情。 6. **questions_survey.xlsx**：问卷调查工具，包含所有问题与耗时统计，按分类体系版本（无引导版与延迟版）进行组织。为便于识别，涉及分类体系介绍与应用的问题以蓝色高亮标注。 ## 匿名化与隐私保护 为遵守隐私法规与企业要求，所有数据均已完成完全匿名化处理： - 企业与软件名称已移除或替换为占位符； - 人口统计学信息及潜在可识别的个人信息已全部删除； - 数据集仅包含非敏感的匿名化定性数据。 ## 使用与引用 本数据集可用于以下场景： 1. 软件工程领域解释需求的二次分析； 2. 需求获取技术的方法学对比研究； 3. 可解释性分类体系的开发与验证； 4. 定性编码与需求工程领域的培训与教学。 若使用本数据集，请引用以下发表成果： > Obaidi, M., Droste, J., Deters, H., Herrmann, M., Ochsner, R., Klünder, J., Schneider, K. (2025). How to Elicit Explainability Requirements? A Comparison of Interviews, Focus Groups, and Surveys. 2025 IEEE 33rd International Requirements Engineering Conference (RE). ## 许可 本数据集采用知识共享署名4.0国际许可协议（Creative Commons Attribution 4.0 International License, CC BY 4.0）发布。 ## 联系方式 若对本数据集有任何疑问，请联系发表成果中列出的通讯作者。