Emotion and Performance Dataset from a Controlled Maintenance Experiment for Emotion-Aware Decision Support Systems (DESDEMONA PRIN 2022 PNRR)

This dataset was collected within the DESDEMONA project (PRIN 2022 PNRR) and contains multimodal data gathered during a controlled laboratory experiment designed to evaluate an emotion-aware Decision Support System (DSS) for maintenance operations. The experimental study was conducted at the University of Calabria in a laboratory-controlled environment using a bottling workstation equipped with an automatic bottle filling machine. Participants were asked to inspect and perform maintenance tasks on the device while specific anomaly conditions were deliberately introduced by the experiment facilitator. The purpose of the experiment was to analyze the relationship between operators’ emotional states, cognitive load, and task performance during maintenance activities. A total of 22 participants (16 males and 6 females, mean age 29.7 ± 7.4 years) took part in the study. The experimental design followed a two-factor experimental framework: Factor A – Workload Density Standard workload condition High workload condition with reduced time and increased task complexity Factor B – Expertise Level Expert participants (trained before the experiment) Novice participants (no prior hands-on training) The combination of these two factors resulted in 44 experimental runs. During each experiment, participants completed a multi-step maintenance procedure involving inspection, anomaly identification, intervention, and verification of system functionality. Multiple sensing and recording systems were used to collect behavioral, physiological, and subjective data. The dataset includes the following types of information: Performance metrics Task Completion Time (TCT) Number of human errors (HEs) Number of attempts (ATs) Video-based data Facial expression and emotion recognition outputs Video recordings of the task execution Physiological data Heart rate measurements collected via wearable devices Self-reported measures Self-Assessment Manikin (SAM) for emotional valence, arousal, and dominance NASA-TLX questionnaire for perceived cognitive workload Additional questionnaires assessing satisfaction, stress, perceived difficulty, and task appropriateness The dataset supports research in human factors, emotion-aware decision support systems, cognitive workload assessment, and human–machine interaction in maintenance environments. All participants provided informed consent prior to participation, and anonymization procedures were applied to ensure data privacy.本数据集依托DESDEMONA项目（PRIN 2022 PNRR）构建，收录了受控实验室实验中采集的多模态数据，该实验旨在评估面向运维作业的情感感知决策支持系统（Decision Support System, DSS）。 本实验于卡拉布里亚大学的受控实验室环境中开展，实验平台为配备全自动瓶装灌装机的灌装工作站。实验过程中，实验组织者刻意设置特定异常工况，要求受试者对设备开展巡检与运维作业。本实验旨在分析运维活动中作业人员的情绪状态、认知负荷与任务绩效之间的关联关系。 本研究共招募22名受试者（男性16名，女性6名，平均年龄29.7±7.4岁）。实验设计采用双因子实验框架： 因子A——工作负荷密度 - 标准工作负荷工况 - 缩减时长且提升任务复杂度的高工作负荷工况 因子B——专业水平 - 专业受试者（实验前接受过培训） - 新手受试者（无实操经验） 上述两个因子的组合共产生44组实验运行批次。 每轮实验中，受试者需完成一套多步骤的运维流程，涵盖设备巡检、异常识别、干预操作及系统功能验证。实验采用多套传感与记录系统采集行为、生理及主观数据。 本数据集涵盖以下几类信息： 1. 绩效指标 - 任务完成时长（Task Completion Time, TCT） - 人为失误次数（Human Errors, HEs） - 尝试次数（Attempts, ATs） 2. 基于视频的多模态数据 - 面部表情与情绪识别结果 - 任务执行过程的视频录像 3. 生理数据 - 可穿戴设备采集的心率测量数据 4. 自我报告类数据 - 用于评估情绪效价、唤醒度与支配度的自我评估模拟人量表（Self-Assessment Manikin, SAM） - 用于评估感知认知负荷的NASA-TLX量表 - 用于评估满意度、压力、感知难度及任务适配性的额外问卷 本数据集可支撑人因工程、情感感知决策支持系统、认知负荷评估及运维场景下人机交互领域的相关研究。 所有受试者均在实验前签署了知情同意书，且数据集已通过匿名化处理以保障数据隐私。