ADAPTS Dataset: Multimodal Data for Context-Aware Workplace Well-Being and Productivity Assessment

The ADAPTS (Adaptive Decision Support System) dataset combines anonymized questionnaire responses collected from 25 voluntary adult participants with synthetically generated environmental, physiological (wearable) data. The synthetic data were created for research, system development, and evaluation purposes, while preserving realistic workplace conditions and participant profiles. This dataset is designed to support research in knowledge graphs, semantic reasoning, digital health, smart workplaces, and context-aware computing. Dataset Structure & Modalities The repository integrates three independent sensing modalities mapping to unique, anonymous user IDs (user_01–user_25): Activity History (simulated): Longitudinal records tracking step counts, personalized goals, active minutes, and sedentary/sitting durations. Health Metrics (simulated): Wearable-derived data covering sleep duration/efficiency, cardiovascular strain, stress indices, and heart rate variability (HRV). Environmental Metrics (populated): IoT-based workplace measurements including air quality ($\text{CO}_2$), thermal comfort (PMV, PPD), and visual comfort (lux, glare index). ADAPTS Queries: A comprehensive Cypher query library for Neo4j containing graph-native retrieval (Level 1) and multi-hop semantic reasoning queries (Level 2) for the ADAPTS Knowledge Graph (AT-KG). Key Components Daily Productivity Score (DPS): A comprehensive metric derived from six weighted components across health, sleep, activity, and environmental comfort factors. ADAPTS Knowledge Graph (AT-KG): A Neo4j-ready framework consisting of 126 semantic nodes and 565 directed relationships. Two-Level Query Framework: Includes Cypher queries for both Level 1 (Direct Semantic Retrieval) and Level 2 (Multi-Hop Semantic Reasoning) to support explainable artificial intelligence (XAI). Anonymization & Privacy All data has been rigorously processed to protect participant privacy. Longitudinal files are tied exclusively to randomized identifiers, and absolutely no personally identifiable information (PII) is included.