SmartQHSE HSE Q&A Open Corpus v2 2026

The SmartQHSE HSE Q&A Open Corpus v2 (2026) is an expanded, curated collection of 64 professional-grade question-and-answer pairs covering the full breadth of occupational health, safety, and environment (HSE) practice. Each entry includes a machine-readable short answer (under 200 characters), a long-form body (1,500–2,500 words), a category label, update date, and slug identifier. This is an expanded version of the original SmartQHSE HSE Q&A Corpus, adding 40 new entries authored in 2026 across eight new topic categories.
Topics covered include: incident rate metrics (TRIR, LTIFR, DART, EMR, severity rate, near-miss ratios, leading vs lagging indicators, frequency rate); ISO 45001:2018 implementation (overview, vs OSHA, certification cost and timeline, clause 6.1, internal auditing, vs ISO 14001); permit-to-work systems (hot work, confined space, LOTO, roles and authorisation, PTW vs JSA); risk assessment (five steps, 5×5 matrix, HIRA vs JSA, dynamic risk assessment, hierarchy of controls); OSHA compliance (300 log, PSM, Focus Four, API RP 754, LOTO, JSA, California SB 553, Form 300A submission, LOPC, CalOSHA heat illness §3395, recordable vs reportable, inspection rights, penalties, General Duty Clause, HazCom GHS); UK HSE (RIDDOR, CDM 2015, COSHH, competent person, improvement vs prohibition notices); GCC regulations (midday work ban, WBGT, OSHAD-SF Abu Dhabi, Saudi MHRSD, Qatar QCDD, UAE MOHRE, Aramco/ADNOC/KOC contractor prequalification); and PPE and miscellaneous topics (hierarchy of controls, manual handling limits, noise action levels, HSE software, NEBOSH IGC).
All content was authored by the SmartQHSE technical writing team and is based on authoritative primary sources including OSHA 29 CFR regulations, ISO standards, BLS survey data, ILO publications, and national ministerial decisions. This corpus is suitable for training domain-specific language models, building HSE chatbots, powering regulatory compliance search systems, and academic research on occupational safety knowledge representation.
Data format is newline-delimited JSON (JSONL). Each line is a self-contained JSON object. Fields: slug (URL-safe identifier), question (the natural-language question), shortAnswer (concise answer under ~250 chars), answer (full long-form body in Markdown), category (topic category slug), updated (ISO 8601 date of last review).