A semantically structured relational database for construction safety analysis

Purpose This study aims to provide experiences of a pilot database development to discuss the opportunities for automation of safety in construction. It also explores how organized data can be used to improve safety analysis and decision-making between and inside projects. Design/methodology/approach The data set included the records of construction works carried out by a Norwegian construction firm, which comprised accidents, events, observations and quality deviations. A normalized relational database containing structured relationships between the affected projects was built, and queries were developed to evaluate safety performance. Findings The systematic database contributed to the information on accidents, projects, stakeholders and risks involved with the construction company. Nevertheless, the nonoverlapping identification numbers, time differences and absence of risk contributing factors might constrain a detailed analysis. Practical implications This systematic database seems to be very helpful in automated safety management regarding trend identification and constraint management. This research promotes the application of data-driven methods in construction activity to improve safety performance and introduce the artificial intelligence-based system in the future. Originality/value This pilot research addresses the experiences of developing a structured database for safety performance analysis from the data received from a construction company. Hence, it demonstrates a pilot model that can accommodate various data types, from accident reports to workforce presence, within a common analysis framework, which can play the role of creating a foundation for automation and data-driven predictive analysis in this field.